ABSTRACT
Title of Document: BILATERAL INTELLIGENCE COOPERATION: THEORY DEVELOPMENT WITHIN THE ‘MISSING DIMENSION’ OF INTERNATIONAL POLITICS
Ryan E. Bock, PhD, 2012
Directed By: Professor George Quester, Department of Government and Politics, University of Maryland-College Park
Bilateral intelligence cooperation has received increased attention in recent years, thanks in part to its publicly acknowledged role in supporting sovereign states in their efforts to counter transnational threats. Until recently most of the scholarship on this topic has been descriptive and atheoretical, with a tendency to treat known examples of cooperation as unique occurrences, rather than as instances of a broader phenomenon. The aversion to theory surrounding bilateral intelligence cooperation has inhibited efforts to develop generalized and contingent explanations about it, such as why it occurs in the first place and under what conditions it is most likely to flourish and atrophy. This dissertation seeks to address these gaps in the Intelligence
Studies literature by leveraging insights from two theoretical traditions from
International Relations—the dyadic democratic peace and relational contracting perspectives—to develop candidate explanations of why certain pairs of states engage in greater degrees of intelligence cooperation over time than others. Two historical case studies—the Anglo-Soviet and Anglo-American military intelligence
relationships during World War II—are presented and analyzed with a view to assessing the relative strengths and weaknesses of each explanation. Through the use of qualitative research methods (i.e., the congruence method and testing of observable implications) and the development of new numeric measurements to capture the depth of intelligence cooperation over time, the author finds that the normative dyadic democratic peace hypothesis and its posited causal mechanism of the regime recognition dynamic are largely corroborated in both cases, thus suggesting that they warrant further consideration as an explanation of the depth of bilateral intelligence cooperation. By contrast, the relational contracting hypothesis and its posited causal mechanism of willful hierarchy are not well supported in either case, thus raising questions about their applicability and generalizability to the larger universe of cases.
BILATERAL INTELLIGENCE COOPERATION: THEORY DEVELOPMENT WITHIN THE ‘MISSING DIMENSION’ OF INTERNATIONAL POLITICS
By
Ryan E. Bock
Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2012
Advisory Committee:
Professor George Quester, Chair Professor Scott Kastner Professor Peter Mallios Professor William Nolte Professor Vladimir Tismăneanu
© Copyright by Ryan E. Bock 2012
Dedication
I would like to dedicate this dissertation to:
• Greta, whose steady patience, unconditional love and quiet encouragement
have sustained me throughout this undertaking.
• Eliot, whose questions and curiosity about the world around him have
provided a constant reminder of why I set out to do this in the first place.
• Gillian, whose compassion and concern for my well-being (“Why is Daddy so
sleepy?”) have helped to put this endeavor into a proper perspective.
ii
Acknowledgements
I wish to thank Professor Miranda Schreurs for providing multiple rounds of early and detailed feedback on a qualitative methods research design (a class project) that ultimately evolved into this dissertation.
In addition, I wish to thank Dr. Michael Warner, Dr. Derek Reveron and Professor
James Igoe Walsh for their insightful comments and helpful guidance on earlier versions of this work. Their collective knowledge of International Relations theory and modern intelligence history—and rare abilities to connect the two together—was an inspiration to me throughout this endeavor.
I also wish to thank Bob O’Hara and the skilled research team of Public Record
Searches for their prompt and professional handling of my requests for scanned copies of public documents from the UK National Archives. Thanks to them, I was able to spend a significantly greater proportion of my time reading and analyzing archival documents back here in the States.
Last but not least, I wish to thank the members of my dissertation committee—
Professors George Quester (Chair), Vladimir Tismăneanu, Scott Kastner, Bill Nolte and Peter Mallios—for their willingness and flexibility to work with me on this project. Of course, any remaining faults, omissions or errors are purely my own.
iii
Table of Contents
Dedication...... ii Acknowledgements...... iii Table of Contents...... iv List of Tables ...... vii List of Figures...... viii List of Abbreviations ...... ix Chapter 1: Introduction...... 1 What’s missing in the ‘missing dimension’ of international politics?...... 1 Research question ...... 3 Structure and implications ...... 5 Terms of Reference...... 7 Intelligence...... 7 Bilateral intelligence cooperation ...... 8 Democracy and autocracy...... 11 Transaction costs...... 15 Anticipated challenges and obstacles...... 18 Outline and structure of study...... 19 Chapter 2: Intelligence Studies Literature Review ...... 22 IR and the neglected colony of Intelligence Studies...... 22 Obstacles surrounding the theorization and academic study of intelligence ...... 24 Why theorize about intelligence?...... 24 Why IR and Intelligence Studies haven’t mixed well (yet)...... 28 Assessing the sub-literature on international intelligence cooperation...... 35 Historical contributions...... 35 Practitioner contributions...... 37 Political science contributions ...... 44 An analytic framework ...... 48 Chapter 3: Theory and Method...... 51 The dyadic democratic peace argument...... 51 Structuralist arguments ...... 53 Normative arguments...... 56 Hypothesis...... 64 The relational contracting argument ...... 67 Origins within the economics literature...... 68 Weber’s anarchic argument ...... 71 Walsh’s willful hierarchy argument...... 73 Hypothesis...... 75 Methodology...... 77 Research objectives...... 77 Case selection criteria ...... 79 Data availability...... 84 Testing and measurement ...... 85 Testing procedures...... 85
iv
Measuring the dependent variable ...... 86 Measuring the independent variables...... 93 Summary...... 98 Chapter 4: Anglo-Soviet Case Study ...... 100 Setting the Scene...... 100 Inauspicious beginnings...... 100 The road to 30 Mission ...... 107 Description of intelligence cooperation machinery ...... 109 Hypothesis Testing...... 113 Rationale for data selection...... 113 Testing the dyadic democratic peace argument ...... 115 Measuring the dependent variable...... 115 Measuring the independent variable ...... 122 Testing for observable implications...... 123 Testing the relational contracting argument ...... 143 Measuring the dependent variable...... 143 Measuring the independent variables...... 146 Testing for observable implications...... 152 Conclusion ...... 159 Chapter 5: Anglo-American Case Study...... 162 Setting the Scene...... 162 World War I: A Beginning ...... 163 The road to the special relationship ...... 170 Description of intelligence cooperation machinery ...... 171 Hypothesis Testing...... 176 Rationale for data selection...... 176 Testing the dyadic democratic peace argument ...... 178 Measuring the dependent variable...... 178 Measuring the independent variable ...... 182 Testing for observable implications...... 183 Testing the relational contracting argument ...... 208 Measuring the dependent variable...... 208 Measuring the independent variables...... 210 Testing for observable implications...... 216 Conclusion ...... 223 Chapter 6: Conclusion...... 226 Argument Summary...... 226 The problem...... 226 Research question ...... 227 Two theories...... 227 Methodology and case selection criteria...... 229 Case findings and theoretical implications ...... 231 Conclusion ...... 237 Appendix I: Analytic Framework for Bilateral Intelligence Cooperation ...... 239 Appendix II: Twenty-Year Mutual Assistance Agreement Between the United Kingdom and the Union of Soviet Socialist Republics (1942)...... 240
v
Appendix III: The Travis-Wenger Agreement of 1942 ...... 244 Appendix IV: The BRUSA Agreement of 1943...... 247 Bibliography ...... 253
vi
List of Tables
Table 1. Bindingness as a Function of Clarity and Formality of Commitment...... 92 Table 2. Anglo-Soviet Liaison Structure and Key Personnel, 1941-1945...... 111 Table 3. Intelligence Goods and Estimated Granularity Measurements...... 119 Table 4. Dyadic Democratic Peace Argument: Anglo-Soviet Case Predictions ..... 124 Table 5. Bindingness Matrix for the Anglo-Soviet WCID, 1941-1945...... 146 Table 6. Asset Inventory for the Anglo-Soviet WCID: British Investments...... 151 Table 7. Relational Contracting Argument: Anglo-Soviet Case Predictions ...... 154 Table 8. Expected Anglo-Soviet Defection-Avoidance Strategies...... 158 Table 9. Anglo-American Liaison Structure and Key Personnel, 1942-1945 ...... 174 Table 10. SLU Presence at Select US Military SIGINT Agencies, 1944-1945 ...... 181 Table 11. Dyadic Democratic Peace Argument: Anglo-American Predictions ...... 184 Table 12. Bindingness Matrix for the Anglo-American WCID, 1940-1945 ...... 209 Table 13. Asset Inventory for the Anglo-American WCID: British Investments ... 212 Table 14. Relational Contracting Argument: Anglo-American Case Predictions... 220 Table 15. Expected Anglo-American Defection-Avoidance Strategies ...... 222
vii
List of Figures
Figure 1. Inter-Governmental Intelligence Cooperation Intensity Scale ...... 8 Figure 2. Depth of Bilateral Intelligence Cooperation (Subcomponents) ...... 11 Figure 3. The Snowing Effect...... 43 Figure 4. Bilateral Intelligence Cooperation Heuristic ...... 63 Figure 5. Regime Type Property Space for WCIDs ...... 66 Figure 6. Weber's Continuum of Cooperative Security Arrangements ...... 90 Figure 7. Organizational Division of Effort for 30 Mission ...... 110 Figure 8. Anglo-Soviet Intelligence Exchanges (Moscow), 1941-1945...... 117 Figure 9. Anglo-Soviet MORR, 1941-1945 ...... 118 Figure 10. Anglo-Soviet AMGIE, 1941-1945 (Face-to-Face Meetings Only)...... 121 Figure 11. Anglo-Soviet AMGIE, 1941-1945 (All Encounters) ...... 122 Figure 12. Organizational Division of Effort for British SLU in the USA...... 176 Figure 13. Anglo-American Intelligence Exchanges (Washington), 1943-1945 ..... 179
viii
List of Abbreviations
AMGIE Average Maximum Granularity of Intelligence Exchanged BLO British Liaison Officer (UK) BRUSA Britain-United States CIA Central Intelligence Agency (US) CIGS Chief of the Imperial General Staff (UK) COMINT Communications Intelligence COW Correlates of War project CSIM Combined Signal Intelligence Meetings DMI Director of Military Intelligence (UK) DNI Directorate of Naval Intelligence (UK) Director of National Intelligence (US) FECB Far East Combined Bureau (UK) FO Foreign Office (UK) GC&CS (GCCS) General Code and Cypher School (UK) G-2 Army Intelligence (US) HUMINT Human Intelligence IGIC Inter-Governmental Intelligence Cooperation IMINT Imagery Intelligence IR International Relations JIC Joint Intelligence Council (UK) MIS Military Intelligence Service (US) MORR Meeting Opportunity Realization Rate NACP National Archives in College Park, Maryland (US) NIE National Intelligence Estimate (US) OB Order of Battle ONI Office of Naval Intelligence (US) OP-20-G Office of Chief Of Naval Operations (OPNAV), 20th Division of the Office of Naval Communications, G Section SGS Soviet General Staff (USSR) SIGINT Signals Intelligence SIS Secret Intelligence Service (UK) Signals Intelligence Service (US) SLU Signals Intelligence Liaison Unit (UK) SSA Signal Security Agency (US) TNA The National Archives in Kew, England (UK) USIC US Intelligence Community (US) USIIB United States International Intelligence Behavior dataset WCID Wartime Cooperative Intelligence Dyad WO War Office (UK) 30 Mission British Military Mission to Moscow (UK)
UK = British acronym US = American acronym USSR = Soviet acronym
ix
Chapter 1: Introduction
What’s missing in the ‘missing dimension’ of international politics?
Bilateral liaison arrangements are a defining characteristic of the intelligence world. …They usually cover a wide range of issues, including the sharing of assessments, raw data, or training facilities and the conduct of joint operations, some of which could lay dormant at any given time. While traces and discussions of particular bilateral relationships can be found in the historical literature, key questions such as "How does such cooperation arise? How frequently and what form does it take? Who authorizes it and arranges it?" are not always easily answered (Lefebvre 2003, 533).
International intelligence cooperation has long played an important behind- the-scenes role in international politics. Over the past century sovereign nation states have formed intelligence alliances against a diverse array of common adversaries, to include international anarchists, the Axis powers, the Soviet Union and transnational criminal and terrorist organizations. That much is evident in the multidisciplinary
Intelligence Studies literature,1 which has shed much light on what Sir Alexander
Cadogan, the distinguished British diplomat, once famously described as the 'missing dimension' of diplomatic history (Andrew & Dilks 1984, 1).
Thanks to this literature, we know much more about international intelligence cooperation than we did in the mid-1970s, when F.W. Winterbotham’s The Ultra
Secret first called the public’s attention to the joint cryptologic activities of the UK and US militaries during World War II (Winterbotham 1974). Historians have illuminated the role of Anglophone intelligence cooperation in securing Allied victory
1 Even 30 years on, Intelligence Studies remains something of a nomad within the academy. As Michael Goodman notes, “intelligence studies is one of those odd disciplines that is comfortable in a variety of academic departments, but perhaps never truly at home in any of them” (Goodman 2006, 58).
1
in World War II (Andrew 1995; Smith 1993; Smith 1996; Alvarez 1999; Budiansky
2000; Warner 2004). Practitioners have highlighted the importance of intelligence sharing arrangements in combating transnational threats and supporting multilateral military operations (Boatner 2000; Lefebvre 2003; Clough 2004; Lander 2004).
Political scientists, for their part, have dissected the inner-workings of intelligence alliances—often with a view to diagnosing inefficiencies, deriving case-specific lessons learned and prescribing organizational remedies (Richelson 1990; Aldrich
2002a, 2004; Bensahel 2006; Byman 2006; Reveron 2006; Sims 2006; Zegart 2007,
2011; Walsh 2010).
Nevertheless, as Lefebvre suggests, deeper questions about bilateral intelligence cooperation remain inadequately explored. What factors compel such arrangements to emerge in the first place? Why are some relationships more vulnerable to cheating and defection than others? And what accounts for varying degrees of cooperation within these relationships? To the extent that these questions are addressed in the Intelligence Studies literature, they have tended to be examined through a descriptive and functionalist (vice analytic and deductive) lens. As a result, our understanding of bilateral intelligence cooperation as a general phenomenon remains somewhat limited and underdeveloped.
While some might argue that theoretical abstractions have no place in the study of such a practical matter as intelligence, a diverse group of intelligence scholars argues otherwise. In an influential essay on this topic, Fry and Hochstein encourage the academic community to move beyond “anecdotally rich or even titillating” accounts of derring-do and connect their scholarship more explicitly to
2
contemporary debates in international politics (Fry and Hochstein 1994, 18). Making more of these connections would, in Svendsen’s view, bring a greater degree of
“explanatory strength” to the study of different aspects of intelligence cooperation
(Svendsen 2009, 709).
At the same time, there are consequences to not connecting intelligence scholarship more explicitly to theory. Gill and Phythian argue that a dearth of
“understandings and explanations of intelligence that transcend particular times and places” poses a significant barrier to society’s larger understanding of intelligence matters, as well as how we think about intelligence and its applications in the present day (Gill and Phythian 2004, 24).
Research question
To those who view intelligence as an instrument of national power, the idea that states would be willing to share such a precious commodity with others—let alone participate in joint or combined intelligence activities—may seem counterintuitive. As it turns out, no state is capable of gathering unilaterally all of the intelligence that it requires. This is especially true of foreign intelligence, given the wide range of potential topics and geographic regions involved. As a coping mechanism, states reach out to foreign counterparts to fill gaps and, in certain cases, reap the benefits of comparative advantage. Thus it becomes possible to reconcile the notions of sovereign nation states simultaneously acquiring foreign intelligence information for selfish reasons (e.g., in order to gain a relative information advantage over their opponents) and sharing some of this information with other states in select instances.
3
A potentially more intriguing question is why certain groups of foreign intelligence services engage in greater levels and degrees of this activity than others.
A cursory review of the past 60 years reveals a diverse set of historical cases that relate in one way or another to this question. Some relationships, such as the US-Iraq and Israel-Palestinian Authority partnerships of the 1980s and mid-1990s, proved to be relatively short-lived and involved the sharing of specific types of sanitized, operational intelligence against common enemies and intelligence targets (National
Security Archive 1983, 7; Gwertzman 1986, A1; Battersby 1996, 1; Horovitz 1996,
9). Other relationships, such as the early 1960s US-Canada foreign intelligence efforts against Cuba, lasted longer and entailed a broader range of foreign intelligence activities, to include mutual tasking of one another’s intelligence assets and the sharing of finished intelligence (Hershberg 2000, 148-155; Munton 2002 and 2009).
These examples suggest that certain pairs of states engage in greater degrees of foreign intelligence cooperation than others. I’m interested in identifying potential factors that account for this variance. In so doing I hope to make a modest contribution to theoretical development in the Intelligence Studies literature, which until recently has been relatively silent on this question. Most explanations of bilateral intelligence cooperation tend to focus on the role of ‘cost-benefit’ and ‘quid pro quo’ calculations of state actors (Richelson 1990; Lefebvre 2003; Clough 2004;
Sims 2006; Munton 2009). I do not dispute that these arguments shed light on how intelligence practitioners approach matters of intelligence cooperation. The problem lies in their conflation of preferences over outcomes with preferences over actions
(Powell 1994, 318). More specifically, these arguments tend to assume that the final
4
outcome of any strategic interaction naturally reflects the original goals of the actors, when in fact it may have been shaped by complex interaction dynamics.2 Such deterministic reasoning often results in truisms (e.g., intelligence services only collaborate with partner services when it is in their interest to do so and when the costs of sharing outweigh the benefits). By approaching this question through the lens of established IR theories, I hope to identify alternative explanations that can more effectively account for varying degrees of foreign intelligence cooperation between states.
Structure and implications
From a practical standpoint, an examination of bilateral foreign intelligence cooperation has much to recommend it. This is seen first of all in the growing profile of foreign intelligence-sharing arrangements in countering transnational terrorism.
Over two decades removed from the end of the Cold War, Washington and its allies have discovered that extant strategic doctrine and traditional instruments of power are insufficient to deter stateless, transnational actors such as al Qaeda, which generally lack a physical “return address against which retaliation can be visited” (Betts 2002,
31). Greater quantities of foreign intelligence—and, by extension, greater degrees of international foreign intelligence cooperation—are required to identify, counter and neutralize these threats.3 Some have gone so far as argue that the effective pursuit of counter-terrorism hinges “largely on the ability of national intelligence services to
2 Frieden refers to this as the sin of commission, which “arise[s] when analysts observe an outcome and draw a direct line from it back to the preferences of the actors.” (Frieden 1999, 52) 3 Derek Reveron attributes the increasing importance of intelligence to a post-Cold War shift in the amount of intelligence required to neutralize a typical threat: “In contrast to the Cold War, where the intelligence-to-force ratio was low, the war on terror requires high levels of intelligence to identify a threat relative to the amount of force required to neutralize it” (Reveron 2006, 455).
5
collaborate with one another effectively in rooting out international terrorist cells”
(Scott and Jackson 2004, 162). The value and substantive contributions of this collaboration is seen in numerous real-world examples. In recent years, foreign intelligence liaison relationships reportedly have supplied information contributing to the killing and capture of al-Qaeda leaders in Pakistan, the disruption of Taliban elements in Afghanistan, and the prevention of multiple transatlantic airline terrorist attacks (Ramesh and Glendinning 2006; Markey 2007; Kulish 2007; Dilanian and
Bennett 2012).
An examination of this topic may also hold implications for how governments such as the United States select and manage their bilateral intelligence alliances against a range of transnational threats, to include terrorism, drug trafficking, human smuggling and money laundering. In the years since the 9/11 attacks, scholars and government officials have debated the relative merits of engaging in counterterrorism cooperation with authoritarian and democratic regimes in the Greater Middle East
(Windsor 2003; Carothers 2003, 2007; Byman 2006; Hayden 2007; Sheridan and
Warrick 2011). In its first National Intelligence Strategy, the US Office of the
Director of National Intelligence (ODNI) expressed a commitment to “forg[ing] relationships with new and incipient democracies” and “establish[ing] new and strengthen[ing] existing foreign intelligence relationships” with “like-minded nations” (ODNI 2005, 8; 15). Skeptics have countered that battle-tested authoritarian partners should not be abandoned in favor of unproven, fragile democratic regimes
(e.g., post-Saddam Iraq) and their nascent intelligence services (Byman 2006, 805-
806). The competing theoretical arguments and tentative findings presented in this
6
dissertation may shed light on the potential risks and advantages of favoring one regime type over the other in matters of international intelligence cooperation.
Terms of Reference
Since intelligence is a topic not often studied in the International Relations
(IR) literature, it will be useful to define several key concepts that appear throughout this study.
Intelligence
A consensus definition of ‘intelligence’ has long eluded academics.
Historians, practitioners and political scientists have tended to supply their own definitions, rather than building upon each other’s contributions (Warner 2002). A central dilemma in these discussions has been whether to conceive of intelligence as an activity, information, process, product, or some combination thereof (Lowenthal
2005, 1-2; Treverton, Jones et al 2005, 7-10; Scott and Jackson 2004, 141). Warner has helped to bridge these disputes by conducting a broad study of existing definitions of intelligence and identifying a core set of attributes. He ultimately concludes that intelligence consists of information, processes and activities that:
• draw to some extent upon secret (classified) material;
• involve government officials carrying out government business;
• concern or are directed against foreign (non-regime) entities;
• result in, contribute to, or entail the production and dissemination of
information for government or military consumers; and
• are geared toward influencing foreign entities without their knowledge.
7
Taking all of these factors into account, Warner argues that an accurate definition of intelligence must contain four essential elements: secrecy, the state, and the twin functions of understanding and influencing the Other. His definition of intelligence – “secret, state activity to understand or influence foreign entities” – proffers a sensible compromise. By conceptualizing intelligence as an umbrella term encompassing multiple types of activities and purposes, the first-order question of
‘what is (and what isn’t) intelligence’ is resolved (Warner 2002).
Bilateral intelligence cooperation
In his study of democratic control and oversight of EU intelligence activities,
Thorsten Wetzling conceptualizes inter-governmental intelligence cooperation as any of a range of activities undertaken jointly by sovereign state foreign intelligence agencies (Wetzling 2005, 11). He locates these activities along a cooperative intensity continuum, illustrated in Figure 1 below:
Figure 1. Inter-Governmental Intelligence Cooperation Intensity Scale
Wetzling’s framework provides a useful starting point for conceptualizing inter-governmental intelligence cooperation. It invokes many of the core attributes
8
from Warner’s definition of intelligence—most notably secrecy, governmental activity, and an emphasis on external (vice domestic) entities. It also attempts to distinguish in scalar fashion between different degrees of intelligence cooperation, which moves closer to addressing some of the intractable questions in the intelligence literature highlighted by Lefebvre at the outset of this chapter.
There are nevertheless several gaps and inconsistencies in Wetzling’s framework. First, it does not meet the standard definition of a continuum: “a coherent whole characterized as a collection, sequence, or progression of values or elements varying by minute degrees” (Continuum 2012). Rather than representing a continuous series of values describing a single concept, his continuum incorporates two distinct concepts. Frequency of contact between intelligence services makes an appearance on the left pole but is absent on the right, where the degree of structural commitment binding these services together is the salient attribute. This is problematic because a given bilateral intelligence relationship could simultaneously exhibit different levels of intensity along the same continuum. For instance, it is conceivable that a formally institutionalized intelligence relationship (‘Advanced’) could generate only sporadic instances of information exchange (‘Little’).
Wetzling’s framework also does not address the possibility that some types of information are more (or less) valuable than others. States weigh carefully the degree of access that they are willing to afford others to their own intelligence. Some partners may be granted access to highly sensitive and specific information that was difficult to acquire, whereas others may receive only summarized information from less sensitive sources. These notional examples illustrate how the concept of
9
intelligence ‘granularity’ must also factor into any examination of the depth of intelligence cooperation between two countries (Clough 2004).4
Finally, Wetzling’s framework does not account for the possibility of one state assuming direct control over another’s intelligence apparatus. More specifically, the framework is anchored on one end in the idea of foreign intelligence services engaging in ad hoc intelligence sharing and on the other end in surrendering their sovereignty to a supranational intelligence apparatus. There appears to be no place in this continuum for a hierarchical relationship in which a dominant state dictates and enforces the rules of intelligence exchange and a subordinate state complies in exchange for various types of remuneration.
In an effort to build upon the foundation of Wetzling’s framework while at the same time addressing its aforementioned problems, I have chosen to frame the depth of bilateral intelligence cooperation – my dependent variable – as a function of three distinct sub-components illustrated in Figure 2: frequency of intelligence exchange, granularity of information exchanged and the degree of bindingness observed in the relationship. The two theories examined in this dissertation speak to different sub- components of the dependent variable. The dyadic democratic peace argument addresses the first two, while the relational contracting perspective addresses the third.
4 In Chapter Four I operationalize granularity as a function of the specificity, sensitivity and original cost of acquiring the information in the first place.
10
Figure 2. Depth of Bilateral Intelligence Cooperation (Subcomponents)
Democracy and autocracy
One of the arguments to be examined in this study is the proposition that democratic intelligence services are generally more cooperative and conciliatory toward one another than they are toward autocratic intelligence services. If such an argument is to be taken seriously, we must be clear on what is meant and implied by the root concepts of ‘democracy’ and ‘autocracy.’ While both terms appear frequently in the Intelligence Studies literature, they are seldom explicitly defined.
This may owe to the fact that so many of the well-known cases involve ‘ideal type’ intelligence services, such as those of the United Kingdom, the United States and the
Soviet Union. When working with such examples, there is a temptation to employ the ‘Potter Stewart standard’ in lieu of offering concrete definitions. Simply put, we will know a democratic or an autocratic intelligence service when we see one.5
5 Christopher Andrew has drawn highly useful distinctions between democratic and authoritarian intelligence systems (Andrew 2004b).
11
The limitations of this approach become apparent when one considers the broader universe of political regimes that conceivably could become involved in bilateral intelligence relationships. Middle Eastern monarchies such as Bahrain and the UAE, for instance, have begun to implement modest procedural reforms that might be characterized as democratic (e.g., holding parliamentary elections and extending the suffrage to women). At the same time, the ruling dynasties in both countries continue to govern in ways that are widely considered autocratic. Both monarchies have preserved the hereditary lines of succession and continue to wield largely unchecked executive powers. Such hybrid regimes do not fit neatly within simple classification schemes. It is therefore necessary to define precisely what is meant by the terms ‘democracy’ and ‘autocracy.’
There is no shortage of definitions of democracy in the American political science literature. In an influential article on this topic, Collier and Levitsky (1997) characterize the range of definitions as follows:
• Classical definitions distinguish in taxonomic fashion among the various subtypes of full-fledged democracies, to include parliamentary, multi-party and presidential systems.
• Procedural definitions define democracy in terms of its day-to-day functions. Perhaps the most well-known procedural definition of democracy is Schumpeter’s.6
• Minimal definitions outline the core set of properties that are common to all viable democracies. Such definitions strive for maximal conceptual validity— i.e., capturing the full range of requisite properties that constitute a democracy in its most minimal sense.
6 Schumpeter’s definition focuses specifically on the method by which leaders are selected and evicted from office: “the democratic method is that institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people's vote" (Schumpeter 1942, 269).
12
• Sub-minimal definitions convey the extent to which a particular regime falls short of being a full-fledged democracy.7
There are likewise numerous definitions of ‘authoritarianism,’ a term that seems initially to have been conceived as an intermediate form of government
(located somewhere between the poles of democracy and totalitarianism) exhibiting circumscribed patterns of political competition, no guiding ideology, an absence of popular mobilization, and a dominant ruler or ruling clique whose powers are ill- defined but whose behaviors are relatively predictable (Linz 1964, as cited in Collier, ed. 1980, 399). In an attempt to account for the diversity of authoritarian regimes, scholars differentiated among military-bureaucratic, organic statist, post-colonial, and neopatrimonial subtypes, among others.
By the late 20th century, authoritarian regimes had ceased to resemble ideal- typical autocracies. Some, such as Poland under General Wojciech Jaruzelski and
Peru under Alberto Fujimori, attempted to maintain their wide-ranging executive powers by making modest political compromises that the leadership deemed necessary for its own survival. Perhaps in a nod to emerging post-Cold War norms of appropriateness, these and other ‘competitive authoritarian’ regimes demonstrated a limited tolerance for political opposition elements but did not hesitate to use “bribery, co-optation, and various forms of ‘legal’ persecution” in an attempt to keep them in check (Levitsky and Way 2002, 59). Thus, despite the fact that domestic opposition
7 Collier and Levitsky refer to these as “diminished subtypes” of democracy. Examples include illiberal democracy, in which leaders are democratically elected but flout constitutional limits on their power and deny basic rights and freedoms to their citizens, and neopatrimonial democracy, in which a single individual wields disproportionate authority over weak democratic institutions (Zakaria 1997; Bratton and van de Walle 1997).
13
elements had been granted limited opportunities for contestation, the strongmen continued to wield executive authority largely as they had before.
These late-20th century examples raise an additional question: how does one distinguish between democracies and autocracies when so many regimes appear to exhibit properties of both? On the one hand, calling these regimes transitional democracies betrays a teleological bias towards democratization. Such a bias is unwarranted, as there are no guarantees that an authoritarian regime that implements a modest package of liberal reforms (e.g., a temporary relaxation of press restrictions) will gravitate toward full-fledged democracy (Carothers 2002, 7). On the other hand, calling these regimes competitive autocracies implies a degree of institutional stability and permanence that isn’t necessarily borne out empirically (Levitsky and
Way 2002).8
A more useful way to differentiate between the concepts of democracy and autocracy is to view them as attributes that may appear simultaneously—and to varying degrees—within the same regime. The Polity Project has employed such an approach in its coding of political regime characteristics from 1800 through 2010.
Each regime is assigned separate annual ‘democracy’ and ‘autocracy’ scores, both on
10-point scales. Full-fledged democracies (i.e., those scoring a 10) are defined as any political system that: allows its citizens to express their preferences about policies and leaders through established institutions and procedures; curbs the exercise of
8 Some authoritarian leaders have undertaken “controlled openings of political space,” only to find themselves ousted by democratic momentum generated by these openings. Such outcomes are not uncommon for competitive authoritarian regimes. Przeworski (1991), for instance, argues that initial compromises made by authoritarian leaders will give rise to popular demands for more substantive and wide-ranging reforms. It is at this point that “moderates” within the government prove unwilling to resort to the old methods of repression to curb dissent. Once these fissures emerge within the leadership, the days of the competitive authoritarian regime are numbered.
14
executive power through institutionalized constraints; and guarantees civil liberties to all of its citizens. Full-fledged autocracies, by contrast, limit the extent of competitive political participation, select their chief executives through a “regularized process of selection within the political elite,” and exercise their power with few institutional constraints. A composite ‘polity’ score is then derived by subtracting the autocracy value from the democracy value (Marshall, Jaggers et al 2011, 14-17).
While the Polity IV dataset is not without its detractors,9 its consistent and theoretically informed coding procedures provide a basis for teasing out the distinctly democratic and autocratic features of hybrid regimes. This in turn enables researchers to place polities in their appropriate ordinal position “along the supposed continuum from tightly closed autocracies to highly open democracies” (Marshall,
Gurr et al 2002, 45).
Transaction costs
A second argument to be explored in this dissertation is the proposition that a pair of highly heterogeneous intelligence services (i.e., serving different political systems and employing different official languages) is more likely than a pair of homogenous intelligence services (i.e., similar political systems and common official language) to engage in highly institutionalized forms of intelligence cooperation.
This argument is rooted in the ‘relational contracting’ branch of economics popularized by Coase (1937) and Williamson (1966, 1979). Relational contracting arguments initially sought to explain a longstanding anomaly of neoclassical
9 See, for example, the critique offered by Munck and Verkuilen (2002)
15
economics: why some groups of utility-maximizing individuals choose to abandon the market (i.e., the price mechanism) and organize themselves into firms.
A central insight of the relational contracting literature pertains to the role of transaction costs in classical market exchange. Transaction costs consist of gathering and interpreting information for purposes of exchange (search costs), writing, negotiating and re-negotiating contracts (negotiation costs), and enforcing the terms of said contracts (enforcement costs) (North and Thomas 1973). Coase argues that entrepreneurs are inclined to create a firm when the costs of doing business in the market exceed what they might be in a hierarchical arrangement. Firms are most likely to form “in cases where a very short term contract would be unsatisfactory” to the entrepreneur, whether because she prefers to avoid the future costs of negotiating and re-negotiating contracts or because, in the face of future uncertainty, she would prefer to ‘lock in’ to a relationship with a known entity (Coase 1937, 392).
Building upon these insights, Williamson argues that buyers and sellers are further compelled to organize themselves into firms when they are unable to withdraw their respective economic investments (e.g., specialized land, labor and physical capital) from the current relationship and redirect them to other activities
(Williamson 1979, 240). More recently, scholars have acknowledged that such asset- specific investments are not limited to traditional factors of production. In the financial sector, for example, investments in social capital (e.g., firms seeking financing from middle-market banks) have proven equally sticky (Uzzi 1999, 501;
Hainmueller and Hiscox 2007, 3).
16
IR scholars have likewise drawn from the insights of relational contracting theory to explain a wide range of phenomena, to include international cooperation, regime formation and the dynamics of state security relationships (Keohane 1982;
Lake 1996; Weber 1997, 2001). More recently, Walsh (2007, 2010) has adapted the relational contracting perspective to his study of intelligence sharing among states.
Such an extension is sensible if one assumes that there is a global intelligence marketplace in which sovereign national intelligence services enter and exit ad hoc.
It is not difficult to envision a scenario in which two of these services—perhaps facing a common intelligence problem or threat—are compelled to abandon the market in favor of a more institutionalized and continuous form of intelligence cooperation with one another. To extend the analogy further, the transaction costs of intelligence exchange can be viewed as the price of conducting ‘arms-length’ ad hoc intelligence exchange in the global marketplace. When two intelligence agencies face particularly high levels of uncertainty, asset specificity, and heterogeneity in their joint efforts against a common intelligence target, they may be inclined to abandon strictly ad hoc modes of intelligence exchange in favor of a more institutionalized arrangement. Consistent with this logic, the degree of institutionalization in a bilateral intelligence sharing arrangement (what Weber describes as the ‘bindingness’ of a security relationship) could be expected to vary in accordance with the expected level of transaction costs borne by the participants. I expand upon this argument and consider its theoretical implications for bilateral intelligence cooperation in Chapter
Three.
17
Anticipated challenges and obstacles
Two challenges loomed large from the outset of this study: the limited availability and suspect reliability of primary source data, and the age of the selected cases. Despite the U.S. Government’s decision to declassify automatically all non- exempted documents over 24 years of age, the fruits of that policy have proven slow to arrive—too slow, as it turns out, for my research objectives.10 I consequently chose to draw upon historical cases for which ample and reliable primary source data are available: declassified US and UK Government documents concerning cooperative military intelligence activities directed against the Axis powers from
1939 through 1945. I accessed these materials at the US National Archives in
College Park (also known as “Archives II”) and the UK National Archives (formerly known as the Public Records Office, or PRO) in Kew, England. Both collections allowed for a thorough examination of Anglo-American wartime intelligence cooperation, as well as a partial examination of Anglo-Soviet cooperation.
A second challenge concerns the applicability of the World War II-era case findings to contemporary bilateral intelligence cooperation. These cases occur amidst a vastly different set of background conditions (namely, conventional war and the rise of single-party states) than those of the present period (i.e., asymmetric war and the rise of stateless, non-hierarchical networks). It could be argued that this divergence in background conditions naturally limits the applicability and
10 The recent development and creation of a United States International Intelligence Behavior (USIIB) dataset offers a potentially promising work-around to the problem of data availability. Thanks to its generation of machine-coded instances of international intelligence cooperation mentioned in wire news releases from 2000 through the present date, scholars may be able to turn to this secondary dataset in the absence of anything similar from primary sources (Aydinli and Tuzuner 2011). I comment on the potential implications of this dataset in Chapter Six.
18
‘generalizability’ of my case findings to the contemporary world. This would be a valid criticism, but I contend that it is overstated on two grounds. First, this dissertation examines the question of whether the presence or absence of certain independent variables shapes the depth of expected bilateral intelligence cooperation against a common threat. That question is arguably just as pertinent and relevant today as it was in the 1940s. Second, the nature of the common threat or intelligence target—whether it be a fascist state, a transnational terrorist group, or even a global pandemic—is arguably of less importance than the fact that the threat exists and that two states see fit to share intelligence against it. So while there are differences between the current era and World War II, I submit that there are sufficient structural similarities to draw limited and typological inferences from the case findings.
Outline and structure of study
A prevailing theme in the Intelligence Studies literature is the notion of government officials carefully weighing the costs, benefits and national security implications of bilateral intelligence cooperation before committing to the exchange of sensitive information with others. While such calculations are undoubtedly present on both sides of a given foreign liaison relationship, it is conceivable that they could be trumped by additional explanatory factors that have yet to be fully considered. It is in this respect that I believe insights from IR could make valuable contributions to the Intelligence Studies literature.
I expand upon this theme in Chapter Two, which serves as the literature review for this dissertation. I argue that the past emphasis on micro-level calculations and cost-benefit reasoning has been a defining feature of the American Intelligence
19
Studies literature, which has come to be dominated by historians, practitioners and political scientists. With several notable exceptions, International Relations theorists have been largely absent from these debates. I contend that this needs to change because IR offers a rich conceptual toolkit with ready applications to the study of intelligence. After considering some of the principal obstacles to (and arguments in favor of) intelligence theorization, I provide a critical review of the existing literature and conclude with the presentation of a crude analytic framework. I then utilize the framework to identify branches of IR theory that can be leveraged to explain bilateral foreign intelligence cooperation.
In Chapter Three I develop candidate arguments from two popular strands of
IR theory: the dyadic democratic peace and relational contracting literatures. The dyadic democratic peace perspective, as adapted to the study of bilateral foreign intelligence cooperation, posits a relationship between the presence of certain regime types within a dyad and the intensity (or degree) of bilateral foreign intelligence cooperation. The relational contracting perspective, by contrast, posits a relationship between intra-dyad heterogeneity and the intensity and form of bilateral foreign intelligence cooperation. The chapter concludes with a discussion of my small-n research design and case selection criteria, followed by an introduction of the two cases that are the focus of the ensuing chapters.
Chapters Four and Five apply the theoretical perspectives introduced in
Chapter Three against the Anglo-Soviet and Anglo-American cases studies. The
Anglo-Soviet case study focuses on the experience of Britain’s 30 Military Mission to
Moscow (also known as 30 Mission) in its dealings with the Soviet General Staff
20
(SGS). The Anglo-American case study focuses on the tenure of the Signals
Intelligence Liaison Unit (SLU) and its interactions with three Washington, D.C.-area
American SIGINT agencies: the Navy’s Op-20-G and the Army’s G-2 Special
Branch and the Signal Security Agency (SSA) at Arlington Hall.
Chapter Six provides a brief restatement of my central arguments and key findings. I use the remainder of this concluding chapter to suggest theoretical implications of my research, highlight some of the contributions this dissertation has made to the scholarly literature on intelligence, and identify future directions for research concerning this topic.
21
Chapter 2: Intelligence Studies Literature Review
IR and the neglected colony of Intelligence Studies
Merriam-Webster’s dictionary defines a colony as “a body of people living in a new territory but retaining ties with the parent state” (Merriam-Webster Online
2012). Metaphorically speaking, this seems an accurate description for the growing number of topical sub-fields within contemporary International Relations. Some of the better-known IR ‘colonies’ include global environmental politics, globalization and international law. Intelligence Studies does not yet represent an IR colony, however. As Fry and Hochstein observe, the traditional view among IR theorists is that Intelligence Studies is “a refinement of diplomatic and military history…[that] is not necessarily significant intellectually” (Fry and Hochstein 1994, 18). If that view is to change, more IR scholars will need to convince their peers that intelligence- related research raises “relevant and interesting questions” for the larger field, demonstrate that IR theory contributes to answering these questions, and argue convincingly that IR-focused intelligence scholarship will “lead to a theoretical core that is sound enough to generate cumulative research and that influences the analysis of international relations more broadly” (Zürn 1998, 620).
I am optimistic that these challenges can be addressed. An important first step will be to identify specific intelligence-related research questions that ‘fit’ with concepts and theoretical debates within International Relations and political science more generally. The question of why certain pairs of intelligence agencies engage in greater degrees of cooperation than others offers a promising start. In addition to
22
holding direct parallels with IR debates over the purpose and scope of state sovereignty, it also touches upon the nature of alliance formation and the structure and form of international regimes and security partnerships. The next step is to extend mainstream IR theories and insights to “aid our understanding of intelligence liaison,” while at the same time relating these findings back to the IR theoretical core
(Svendsen 2009, 708). I seek to do both in this dissertation.
Given the multidisciplinary nature of contemporary Intelligence Studies, it is outside the scope of this study to provide a comprehensive literature review of the entire sub-field. This chapter seeks instead to do three things. First, it offers a rationale for theorizing about intelligence and identifies some of the principal obstacles surrounding the theorization and academic study of intelligence. This discussion serves to highlight the role and significance of theory in Intelligence
Studies, as well as identify some of the structural, historical and cultural barriers that have made intelligence an unattractive topic to so many American political scientists over the years. I find that these barriers are not as daunting as they once were, thanks to recent geopolitical developments, a more liberal U.S. declassification policy, an evolution in the academy’s perceptions of intelligence, and a rehabilitation of qualitative research methodologies among IR scholars.
This discussion sets the stage for the second objective of the chapter: a critical assessment of the relatively sparse and atheoretical literature on international intelligence cooperation, accompanied by a closer look at its limited treatment by political scientists. The chapter concludes with the presentation of a crude analytic framework. The purpose of this framework is to identify existing branches of social
23
science theory that have potential applications to the study of international intelligence cooperation.
Obstacles surrounding the theorization and academic study of intelligence
Why theorize about intelligence?
One IR scholar defines theory as a “disciplined process of world construction, whereby a perspective [is] first elaborated in ideal-typical fashion and then used as the baseline from which to rigorously produce an account” (Jackson 2004, 31-32).11
A skeptic might counter that such abstractions play no useful role in the study of intelligence because they cannot possibly capture the manifold complexities and nuances of buzzing, blooming ‘intelligence reality.’ Those who view theories as being synonymous with covering laws would probably be sympathetic to such an argument. But for those who view theory as a means of providing generalized and contingent explanations about particular phenomena (not to mention distilling, organizing and making sense of said phenomena), the skeptic’s position fails to convince.
As a CIA analyst observed nearly 40 years ago in an internal Agency publication, theory imparts cumulative coherence to real-world experiences and insights from the intelligence profession that might otherwise “persist only as an undifferentiated mass of data and impressions” (Mitelman 1974). It does so by forcing analysts to distinguish between drivers (independent variables), outcomes
(dependent variables) and linchpins (operating assumptions) of intelligence-related
11 To this, most neo-positivists would probably add that such “accounts” must then be tested against new data (i.e., independent of the initial observations used to construct the baseline).
24
phenomena (Gill and Phythian 2004, 3; Davis 1999). Theory also imposes analytic discipline and rigor on a field of study that to date has consisted largely of ad hoc historical accounts and ‘story-telling’ (Fry and Hochstein 1994, 15; Gill and Phythian
2004, 24).
As noted previously, much of the academic literature on intelligence has been atheoretical in orientation. This is reflected in classical intelligence texts, edited volumes on the practice and history of intelligence, and in the pages of the leading peer-reviewed intelligence journals.12 Historians tend to treat each intelligence- related episode as a unique event unto itself, rather than as a single instance of a broader class of events. Current and former intelligence practitioners, for their part, seek to distill and articulate practical insights gleaned from their years in the field— often with a view to highlighting ‘what works’ and what does not. Rounding out the field is a small group of political scientists whose scholarship has been characterized as “mainly empirical and analytical-descriptive” (Fry and Hochstein 1994, 15). Much of the latter group’s work could be described as a mixture of contemporary history and traditional social science, in the sense that it employs a combination of traditional case study techniques,13 crude schematics (e.g., ‘the intelligence cycle’14) and
12 ‘Classical’ studies of intelligence include Roberta Wohlstetter’s Pear Harbor: Warning and Decision, Harry Howe Ransom’s The Intelligence Establishment and Christopher Andrew’s For the President’s Eyes Only. Some of the more recent edited volumes on intelligence include: Alvarez (1999), Stafford and Jeffreys-Jones (2000), and Johnson (2007). Intelligence and National Security, The International Journal of Intelligence and Counterintelligence, and the CIA’s Studies in Intelligence arguably constitute the ‘flagship’ journals devoted to the study of intelligence. Each caters to a mixed audience of intelligence scholars and practitioners. 13 These in-depth case studies have addressed a number of contemporary intelligence topics of interest, to include: intelligence oversight (Johnson 1985), reform (Zegart 2007), and politicization (Cahn 1998). 14 The ‘intelligence cycle’ is a conceptual model that is often used to describe the process by which intelligence functions in service of the nation state. In the United States, the intelligence cycle is generally depicted as a feedback loop consisting of five sequential steps: (1) planning and direction;
25
theoretical frameworks (e.g., bureaucratic politics) to buttress largely inductive arguments about particular aspects of intelligence.
These publications have been instrumental in illuminating ‘first principles’ of intelligence, such as what it is (and what it is not), how it is used, why it is important, and what its principal limitations are. But as Gill and Phythian argue, they have been arguably less successful in contributing to our understanding of the intelligence gestalt (Gill and Phythian 2004, 25). Rather than constructing and testing hypotheses, these authors have focused primarily on highlighting the activities and organization of the US Intelligence Community (USIC), describing how certain intelligence processes work, conducting ‘post-mortems’ on notable intelligence failures and successes, identifying strengths and weaknesses of the USIC, and proposing organizational remedies.15
Fortunately, intelligence scholars are beginning to acknowledge the theoretical void in Intelligence Studies. They are nevertheless of mixed views on what should be done to fill it. This was evident during a 2005 symposium16 in which a group of 40 distinguished intelligence scholars and professionals could not reach agreement on the definition of intelligence; the nature, purpose, feasibility, and desirability of a theory of intelligence; and appropriate starting points for theory
(2) collection; (3) processing and exploitation; (4) analysis and production; and (5) dissemination. “The Intelligence Process.” 2008.
26
development.17 Part of the problem may have owed to the title of the symposium:
Toward a Theory of Intelligence. The notion that a single, unifying theory of intelligence is both possible and desirable is problematic. For one thing, the pursuit of ‘theories of everything’ has bedeviled scholars in far more mature disciplines (e.g., mathematics and physics). For another, such a pursuit is likely to remain hamstrung by the dizzying array of analytic puzzles, research agendas, normative commitments, levels of analysis, and entry-level assumptions that drive contemporary intelligence scholarship.18 An arguably more practical and attainable approach, suggested by Gill and Phythian (2004), is to:
be realistic and accept that there can be no over-arching universal ‘theory of intelligence’ because of complexity and the inevitability of the social construction of knowledge. What we can strive for is greater shared understanding of the central questions to be asked about intelligence and what items from the conceptual toolkit will be useful and/or appropriate for the analysis of the phenomenon at the various social levels at which it operates (25).
If intelligence is under-theorized and in need of a more robust ‘conceptual toolkit,’ political science would appear to have much to contribute in the way of support. The political science sub-discipline of International Relations, in particular, offers numerous explanatory paradigms that could be leveraged against intelligence- related research questions and phenomena (Svendsen 2009, 727). Recent edited
17 According to the final workshop report, “[t]he participants were divided over whether a theory of intelligence could be developed in some form and asserted that at best a theory of intelligence is ambiguous” (Treverton et al 2005, 10). 18 Intelligence scholars at the seminar were divided, for example, on whether the elusive theory of intelligence should be normative (describing an ‘ideal type’ of intelligence system or community against which imperfect practical examples could be compared) or empirical (explaining the variance of certain intelligence-related outcomes, such as decision-making). There was also disagreement on whether to focus analysis at the level of individuals, organizations, states, or the international system.
27
volumes such as Gill, Marrin and Phythian’s Intelligence Theory (2009) suggest the beginnings of heightened interest among IR scholars in theoretical treatments of intelligence. Nevertheless, apart from this work and several other notable exceptions,19 the canvas has remained largely blank. So, wherein lies the problem?
More to the point, why are so many political scientists and IR scholars reluctant to study intelligence?
Why IR and Intelligence Studies haven’t mixed well (yet)
Among the academic disciplines that comprise contemporary Intelligence
Studies, political science would appear to be especially well-equipped to tackle deeper theoretical questions. Indeed, the topic of bilateral intelligence cooperation is analogous to cooperative phenomena studied in mainstream IR, such as alliance formation and behavior (Walt 1987, Snyder 1990, Christensen and Snyder 1990, Lai and Reiter 2000, Gibler and Rider 2004, Owen 2005). IR also offers a rich catalogue of conceptual paradigms and methodological approaches that conceivably could be leveraged against this topic, to include neo-realism, neoliberal institutionalism, constructivism, game theoretical treatments and rational choice perspectives. These parallels notwithstanding, American IR scholars have exhibited a longstanding aversion to the study of intelligence. It is difficult to pinpoint precisely why this is so,
19 In her path-breaking study of the Pearl Harbor attack, Wohlstetter (1962) drew upon principles of communications theory to explain why, despite advance warning of a Japanese attack, the US was caught by surprise. Jervis (1968, 1976) and Tetlock and Maguire (1985) have leveraged various strands of cognitive theory to explain why decision-makers sometimes process information (including intelligence) in a way that avoids value trade-offs and coheres with pre-existing images and belief structures. And more recently, Sims (2006) and Walsh (2007) have employed neo-realist and transactional cost economics arguments to explain the conditions that shape and underpin foreign intelligence liaison partnerships.
28
but four inter-related factors warrant our consideration: conflicting value systems, historical suspicions, inadequate sources and methodological commitments.
Conflicting value systems. IR scholars tend to privilege the traditional tools of statecraft (e.g., weapons, trade and foreign assistance) in their study of international politics. They generally pay less attention to ‘non-traditional’ or controversial instruments of state power, such as surveillance and covert action
(Anderson 1998/1999, 404, as cited in Scott and Jackson 2004, 142).20 Loch Johnson suggests that this may owe to a visceral aversion among political scientists to unsavory subjects: “…[T]hey often say, ‘Well, I don’t want to study something as nasty as intelligence: these agencies are overthrowing governments and stealing secrets, and sometimes even trying to kill people. I just don’t think it’s a very good topic to study’” (Johnson 2007, 423). According to this logic, it is the unpleasant nature of intelligence—part of what Max Weber once called “the “morally dubious means or at least dangerous ones” employed by the state—that renders it unfit for scholarly examination (Weber 1946).
This rationale is puzzling, given the robust political science literature that already exists on far more violent and unpleasant phenomena in international politics, such as great power wars, military intervention, genocide and state terror. If it is not the inherent ‘nastiness’ of the topic that bothers political scientists, then it must be some other set of factors. Two that readily stand out are the conflicting value systems
20 Between 2002 and 2012, only three articles concerning intelligence appeared in the leading peer-reviewed scholarly journals of American International Relations. This finding is based upon an electronic database query conducted on all non-book review articles that: (a) contained the term “intelligence” in the title; and (b) were published in any of the following peer-reviewed journals: American Political Science Review, International Organization, International Security , or World Politics. The only results retrieved from this query were: Zegart 2005, Rovner, Long et al 2006 and Powell 2007.
29
and entrenched suspicions of the intelligence community that date back to the early
Cold War period. The potential for a values clash is seen first of all in the contrasting missions of the academy and the intelligence community. At the risk of oversimplifying: whereas the academy privileges the open pursuit and public dissemination of knowledge in the service of humankind, intelligence organizations
— acting on behalf of sovereign state governments — privilege the acquisition and protection of information (both through clandestine and open means) while at the same time seeking to influence others. It is arguably this combination of secrecy and influence that represents the primary ethical dilemma for political scientists. How does one undertake an objective study of the state and its secret activities without becoming complicit in (or being manipulated by) them? According to one scholar, the subtle risks of manipulation — and accordingly, contamination of one’s own research — are real and ever-present for anyone who studies intelligence:
The CIA is not an ordinary government agency; it is an espionage agency and the practices of espionage … are diametrically opposed to those of scholarship. Scholarship is supposed to favor objective analysis and open discussion. The close relationship between intelligence agencies and scholars thus poses a conflict of interest” (Gibbs 2001).
Historical suspicions. Concerns about secrecy and manipulation first came to a head in the mid-1960s, when Ramparts magazine, The New York Times and The
Washington Post exposed the CIA’s clandestine funding of U.S. students, academics and private organizations dating back to the early 1950s.21 A more comprehensive
21 In fact, the relationship between the U.S. intelligence establishment and the academy predates the Cold War. The Office of Strategic Services (OSS), for example, employed a number of university professors to staff the Central European section of its fledgling Research and Analysis Branch. This diverse group of economists, historians, sociologists and German intellectuals supported the Allied war effort by assessing day-to-day developments in Nazi Germany and, later, attempting to anticipate some of the challenges of postwar occupation (Katz 1987, 446-448).
30
and official accounting of these activities surfaced in the 1976 Church Committee
Reports,22 which chronicled the CIA’s covert funding of the National Student
Association from 1952 through 1967 and called attention to its continued operational use of American academics in “over 100 American colleges, universities, and related research institutions.”23 According to the committee, these scholars were used for a range of intelligence purposes, to include making introductions, collecting foreign intelligence and “publishing books and other material to be used for propaganda purposes abroad” (U.S. Senate 1976, 185-190).
Against this historical backdrop, it is not difficult to understand why scholars might be reluctant to study or associate themselves with intelligence organizations— particularly when they fear the possibility of becoming accomplices in or apologists for the very activities that they purport to examine objectively. While the U.S.
Intelligence Community has worked to repair its relationships with American universities (e.g., through open awarding of competitive research grants, overt participation in employer recruitment events, and increased sponsorship of scholars- in-residence programs), the perceived risks of academic manipulation and its implications for one’s scholastic integrity may continue to deter some from studying intelligence.
22 The Church Committee is the informal name for the United States Senate Select Committee to Study Governmental Operations with Respect to Intelligence Activities, chaired by the late senator Frank Church (D-ID). The committee was established in 1975 to investigate the legality of intelligence activities conducted by the CIA and FBI, among other government agencies. While evidence of some of these activities had been exposed earlier in a series of investigative journal articles, the Church Committee’s hearings and ultimate findings are generally credited with giving impetus to the intelligence oversight reforms of the late 1970s. 23 The committee defined “operational use” of individuals as the “recruitment, use, or training, on either a witting or unwitting basis” for involvement in “covert action, clandestine intelligence collection (espionage) and various kinds of support functions” (U.S. Senate 1976, 184).
31
Insufficient and inadequate data. Academics may also be reluctant to study intelligence because it is a topic for which very little primary source information is available. The fact that so much of what is available is supplied selectively—and perhaps self-servingly—by governments may pose an additional deterrent (Jervis
1986, 143; Aldrich 2002b, 148).24 Faced with such daunting research obstacles, many intelligence scholars have turned to alternative sources, to include interviews with former officials, retiree memoirs and even their own first-hand experiences.
Reliance on these sources places the state and its agents (the objects of study for many political scientists) in the position of information gatekeeper. Since the provenance and accuracy of this information can seldom be independently corroborated, scholars who rely upon it are vulnerable to the charge that they are
“legitimizing and perpetuating the ideology of the state” (Scott and Jackson 2004,
152). Such accusations may have particular resonance among political scientists— particularly those who recall similar charges leveled by critics of the ‘Bringing the
State Back In’ movement of the late 1970s and 1980s in Comparative Politics
(Abrams 1977; Cammack 1989).
Methodological commitments. In the absence of robust and independently acquired intelligence datasets, much of the early intelligence scholarship took the form of intensive small-n case studies.25 This methodological approach, described by
Harry Eckstein as ‘configurative-ideographic,’26 came under fire during the 1990s
24 Aldrich argues that, by relying on declassified intelligence documents as their primary (and in some cases, sole) source of information, intelligence scholars unwittingly limit their research to a “pre-selected menu” of the state’s choosing (Aldrich 2002b, 148). 25 Two well-known examples include Ransom (1970) and Johnson (1985). 26 Eckstein elaborates: such case studies contain a “configurative element [that] aim[s] to present depictions of the overall Gestalt (i.e., configuration) of individuals: polities, parties, party
32
from proponents of more rigorous and scientific methods of inquiry (Keohane, King and Verba 1994; Bates 1997; Martin 1999).27 While these methodological criticisms were not leveled explicitly at intelligence scholars, it is certainly conceivable that they have had an effect on the number of political scientists who study intelligence, as well as where the subject of intelligence is taught within the academy.28
* * * * * * * * * * *
Recent developments suggest that each of the aforementioned obstacles to the study of intelligence is beginning to recede. In the aftermath of the 9/11 attacks,
American universities have expanded their curricula to address topics that previously had not received much scholarly attention in their own right, such as transnational terrorism, radical Islam and intelligence.29 Vocationally-oriented departments have also begun to offer coursework and degree programs that focus specifically on the history and conduct of intelligence (May 1995; Goodman 2004).30 Political
systems, and so on” and an “idiographic element [that] …either allow[s] facts to speak for themselves or bring[s] out their significance by largely intuitive interpretation, claiming validity on the ground that intensive study and empathetic feel for cases provide authoritative insights into them” (Eckstein 1992, 136). 27 A defining text of this movement is Designing Social Inquiry, whose authors argue that all social science research, whether quantitative or qualitative in nature, must by definition exhibit the following four characteristics: a shared goal of inference beyond the immediate data; the use of transparent and explicitly codified testing procedures; conclusions couched within a “reasonable estimate of uncertainty”; and a set of rules on which a given argument’s validity depends (Keohane, King and Verba 1994, 7-9). 28 At least initially, these criticisms were directed primarily at area studies specialists. For a crystallization of the key arguments on both sides of this debate, see Bates (1997) and Johnson (1997). 29 This has been fueled to a certain extent by U.S. ‘academic centers of excellence’ (CAE) programs such as those sponsored by the National Security Agency and the ODNI. CAEs typically receive funding, resources and post-graduate recruiting opportunities in exchange for their development of curricula that promote professional competencies defined by the sponsoring organization. 30 Some of these programs include: the Institute for Intelligence Studies at Mercyhurst College (Erie, PA), the University of Georgia’s School of Public and International Affairs (Athens, GA), the University of Maryland’s School of Public Policy (College Park, MD), and Tennessee State University’s Center for Intelligence Studies (Nashville, TN).
33
scientists teach many of these courses. Meanwhile, the Intelligence Studies Section of the International Studies Association (ISA) continues to provide a forum for intelligence scholars across multiple disciplines to share and discuss their ideas on the practice and theory of intelligence.
There have also been positive developments on the methodological front.
Proponents of qualitative methods have argued for the continued use of case studies, process-tracing, fuzzy-set logic and the congruence method. They contend that these techniques can be leveraged effectively against policy-relevant small-n research questions, provided that researchers adhere to broadly accepted standards of scientific inquiry and rigor (George and Bennett 2005; Brady, Collier and Seawright 2004;
Bennett and Elman 2006). There are also encouraging signs that political scientists are beginning to find creative uses for some of these methods in their treatment of intelligence-related topics.31
There are likewise preliminary indications that data availability—perhaps the most intractable obstacle—may be beginning to recede as well. This is seen in the development and creation of the aforementioned USIIB dataset, which could serve as a powerful tool for quantifying international intelligence cooperation, as manifested in secondary source data (Aydinli and Tuzuner 2011). It is also seen in the recent efforts of the US Government to liberalize its once-dilatory declassification regime.
All non-exempted secret documents older than 24 years of age are now automatically
31 See, for example, two papers presented at the ISA’s 49th annual convention: "Reforming U.S. Intelligence: When and How Should Intelligence Be Centralized?" (Durbin 2008); and "Counterinsurgency Cooperation: The Importance and Structure of Intelligence Sharing” (Walsh 2008).
34
declassified.32 It remains to be seen what impacts these exemptions will have on the pace, quantity and topical diversity of released materials. At the very least, however, the new policy appears to bode well for future historical and political research on
American intelligence matters (Shane 2006a).
Assessing the sub-literature on international intelligence cooperation
There is a small but distinct ‘sub-literature’ within Intelligence Studies devoted specifically to the subject of foreign intelligence liaison.33 This sub-literature consists primarily of historical studies from the Second World War and the early Cold
War period, ‘lessons learned’ from current and retired intelligence professionals, and what might best be described as ‘stock-taking’ publications that highlight the risks, benefits, challenges and strategic significance of foreign intelligence cooperation in the contemporary era. Only recently have IR theorists begun to address the topic of foreign intelligence liaison.
Historical contributions
The historical literature on international intelligence cooperation defies easy characterization. The sheer diversity of perspectives and sources has contributed to an uneven landscape of accounts (Charles 2000, 259). A common focus of many of these publications concerns Anglophone intelligence links during World War II and the early Cold War. The disproportionate academic coverage of these relationships
32 This occurred in spite of the Clinton and Bush Administrations’ reported reclassification of thousands of historical intelligence community documents at the U.S. National Archives from 1999 through at least 2006 (Shane 2006a). 33 ‘Foreign intelligence liaison’ is a term often used by intelligence practitioners. It refers to the various government-to-government arrangements for sharing secret intelligence.
35
most likely owes to their historical uniqueness, as well as to the increasing availability of information about them—whether in the form of oral histories, first- person narratives, archival information or declassified in-house studies. The gradual declassification and release of these sensitive materials is perhaps attributable to the democratic political systems of the Anglophone governments, each of which has sought to balance the needs of national security with those of open government.
Some of the initial historical accounts of World War II-era Anglophone intelligence sharing, such as The Ultra Secret (Winterbotham 1974) and A Man
Called Intrepid (Stevenson 1976), were written primarily from a first-person participant perspective. An oft-cited weakness of these publications was their tendency to dramatize, if not embellish or misrepresent, the contributions and achievements of the protagonists (Charles 2000, 260). More detached and authoritative accounts, such as Bradley Smith’s The Ultra-Magic Deals and the Most
Secret Special Relationship, 1940-1946 and edited volumes such as David Alvarez’s
Allied and Axis Signals Intelligence in World War II, did not appear until the 1990s.
That decade marked an unprecedented, albeit limited, opening of government archives. This archival glasnost was driven largely by London’s Open Government
Initiative of 1993 and a Clinton Administration Executive Order (EO 12958) issued two years later.34 These initiatives resulted in the declassification and release of
34 The UK’s Open Government project was launched in 1993. Also known as the Waldegrave Initiative (so named for its chief sponsor, Civil Service minister William Waldegrave), it called upon the British Government to adhere to three themes in its control of official information: “handling information in a way which promotes informed policy-making and debate; providing timely and accessible information to the citizen to explain the Government's policies, actions and decisions; and restricting access to information only where there are good reasons for doing so” (Norton-Taylor 1993, 2). The Waldegrave Initiative has led, among other things, to the annual declassification and release of national security and intelligence-related documents through the auspices of the National Archives in Kew, England. US Executive Order 12958, published in 1995, called for the declassification of non-
36
archival materials in both countries from the pre-1945 period, thereby affording historians new windows into some of the key Allied decisions taken during the
Second World War (Aldrich 2002b, 135).
While new materials from this period continue to be released, there are limits to both countries’ archival largesse—particularly when sensitive matters of foreign intelligence liaison are concerned. Neither the UK nor the US Government seems particularly keen to divulge details of whatever foreign intelligence-sharing relationships may or may not have persisted after V-J day. In this regard, Smith’s observation from over 15 years ago remains apropos: “In nearly every archive one chooses to explore, a very restrictive policy immediately shows itself as soon as one seeks to look at events that occurred in September 1945 or the following months and years” (Smith 1993, xi).35
Practitioner contributions
In the pages of professional journals such as Intelligence and National
Security and the CIA’s Studies in Intelligence, current and retired intelligence professionals have reflected on their own experiences in the realm of foreign intelligence liaison. This practitioner literature provides rich insights into the everyday mechanics of foreign intelligence liaison, as well as its perceived benefits, challenges and risks. A brief review of these insights will help to inform our
exempted national security information dated more than 25 years old. It is probably no coincidence that both initiatives emerged in the immediate aftermath of the Cold War. In fact, the Clinton Administration specifically highlighted the “dramatic changes” of this period (e.g., the disintegration of the Soviet Union and the reunification of Germany) as catalysts for EO 12958. 35 As recently as 2002, Aldrich echoed these sentiments in his lament of the “constabulary” pace at which the UK and US Archives declassify intelligence records from the Cold War (Aldrich 2002, 147).
37
understanding of international intelligence cooperation more broadly, as well as highlight basic principles that constitute the ‘conventional wisdom’ surrounding it.
For purposes of simplicity, I find it useful to divide the practitioners’ insights into two over-arching thematic categories: incentives to cooperate and additional considerations.
Incentives to cooperate. From an intelligence practitioner’s perspective, the decision to engage in intelligence cooperation with a foreign partner may be motivated by a variety of considerations. Perhaps the most obvious of these is selfish operational need. As Sims observes, states sometimes struggle to realign their collection systems to target rapidly emerging threats in ‘denied’ areas.36 Leveraging a trusted partner’s intelligence assets in the short term (usually in exchange for some other intelligence good) can increase the recipient’s access to these areas, thereby increasing the timeliness of information flow to key decision-makers and avoiding the short-term risks and costs associated with a unilateral ‘ramp-up’ of intelligence collection (Sims 2006, 206).
There are also cost-efficiency incentives for undertaking longer-term bilateral intelligence cooperation. As several practitioners have noted, few states are in a position to collect, process, and analyze efficiently all of the intelligence that they require. For this reason, a state might choose to cooperate with trusted partners that have access to geographic “areas… where its own reach is limited” (Lander 2004,
490). In the long run, such geographic divisions of effort can promote intelligence economies of scale, thereby reducing operational costs for each participant and
36 A denied area refers to any geographic region or locale against which a particular agency is unable to collect intelligence.
38
augmenting the overall intelligence effort against commonly perceived threats, such as narcotics trafficking, human smuggling and terrorism (Lefebvre 2003, 534;
Richelson 1990, 311; Boatner 2000, 88).
A foreign intelligence service may also pursue bilateral cooperation as a means of extracting rents. For instance, a service with highly sought-after geographic access may be more inclined to cooperate in intelligence matters if its prospective partner is a well-funded, technologically advanced intelligence service that offers the promise of exposure to specialized skills and cutting-edge technologies that would otherwise prove difficult to acquire. Rents may also take the form of non-intelligence goods, to include any number of economic, political and military benefits (Sims 2006,
197). As noted below, these types of ‘asymmetric’ intelligence exchanges can present troublesome implications for the recipient—particularly when the provider is an authoritarian regime.
States also have come to view intelligence cooperation as an alternate inter- state communication channel, as distinct from the traditional mechanisms of government-to-government and military-to-military exchange (Lander 2004, 482).
An example of this ‘intelligence diplomacy’ was seen during the implementation phase of the 1998 Wye River Peace accords, when the CIA was called upon to monitor the Palestinians’ domestic terror crackdown and facilitate a trilateral intelligence dialogue with the Palestinian and Israeli security services (Tenet 1998,
A23). It was hoped at the time that the professionalism of the intelligence services and their mutual respect for one another would enable them to tackle politically controversial problems that previously had bedeviled the politicians and diplomats.
39
Another variant of intelligence diplomacy is seen in the Trojan Horse analogy.
In this scenario, one government leans upon its intelligence services to forge cooperative ties with its counterparts in another country. While the relationship is initially presented as a means of working together on mutual problems of interest, the initiator has a more strategic objective in mind: laying the groundwork for an
‘entangling alliance’ that binds the two governments together against a common threat or adversary. Christopher Andrew has suggested that the UK harbored such a motive during its cultivation of intelligence ties with the United States during the late
1930s and early 1940s. Eager to draw the ostensibly neutral United States into the war effort against Hitler, London saw an enabler in the person of William J.
Donovan, a close FDR associate with strong pro-British sympathies. Andrew has gone so far as to argue that “the primary British intelligence objective in the United
States during the spring and summer of 1941 was to create an Anglo-American intelligence alliance with Donovan as U.S. intelligence coordinator” (Andrew 1995,
98).
Additional risks and considerations. Practitioners are careful to note that, for each potential benefit conferred by foreign intelligence liaison, there is always an accompanying risk. The most commonly cited risk involves the disclosure of sensitive information to unauthorized third parties. There is a common perception among intelligence practitioners (or at least, among those who have written publicly on the subject) that the risk of unauthorized disclosure is compounded within multilateral arrangements, where the sharing procedures become more complex and the handling procedures tend to be more lax, as in NATO (Lefebvre 2003, 532;
40
Clough 2004, 602). When sensitive information is leaked outside authorized channels, the damages can range from international embarrassment to circular intelligence reporting37 to the compromise of sources and methods used to gather intelligence (Richelson 1990, 318).
An additional consideration relates to the issue of ‘granularity,’ or what
Clough defines as “the degree of [one party’s] access to [another’s] package of intelligence” (Clough 2004, 603). States are understandably discriminating in their selection of what types of intelligence they share with others. The degree of access therefore ranges from the highly general to the highly specific. While practitioners do not dismiss out of hand the potential value of low-granularity intelligence goods (e.g., finished intelligence assessments), they are generally keener to acquire high- granularity goods (e.g., ‘raw’ and single-source reporting) because of the potential for increased precision, accuracy and proximity to the objects of collection (Lander 2004,
492). Whether a government is willing to share highly granular information is thought to be dependent upon the structure and formality of the bilateral intelligence relationship. For the sharing of highly sensitive intelligence, governments seem to privilege formalized bilateral arrangements because more effective controls can be exercised over the dissemination, handling and protection of information (Clough
2004, 603).
A third consideration involves the risk of being deceived and manipulated by a foreign partner with a hidden agenda. In agreeing to share intelligence with another intelligence service, one government (the recipient) assumes the risk that its
37 A well-known example of circular intelligence reporting is the Curveball episode, which is described in further detail in Note 39 on page 44.
41
counterpart (the provider) will deliberately and selectively supply (or even fabricate) information with a view to influencing the recipient’s foreign policies. If one accepts
Warner’s definition of intelligence (‘secret, state activity to understand or influence foreign entities’), it becomes evident that this risk is present in all types of intelligence liaison relationships, regardless of the regime types involved.
Nevertheless, it could be argued that the risk is particularly high (and perhaps highest) in relationships involving an intelligence service of an authoritarian regime. I submit that an intelligence recipient’s risk of being ‘snowed’ by an authoritarian provider is theoretically greatest when the following three conditions are present in a liaison relationship:
1) The recipient specifies up front its desire for certain types of intelligence against a particular target. 2) The authoritarian regime has its own agenda (independent of and separate from the recipient’s) to pursue vis-à-vis said target. 3) In exchange for its cooperation, the authoritarian regime receives an uninterrupted flow of intelligence and non-intelligence rents from the recipient.38
This phenomenon, which I call the ‘snowing effect,’ may hold implications for bilateral intelligence sharing. As illustrated in Figure 3, transactional intelligence sharing arrangements involving authoritarian regimes have the potential to degenerate into a vicious cycle. This is seen in the flow of non-intelligence rents — in particular, goods such as military and economic assistance that are supplied outside of intelligence channels — that contribute directly to the survival of the authoritarian regime and the ruling clique. I argue that this flow of goods can create a powerful
38 All three conditions are conceivably present in the USIC’s current intelligence relationships with authoritarian regimes of the Greater Middle East. The second condition is particularly relevant in the context of the current counter-terrorism campaign, given that some of Washington’s counter- terrorist allies (e.g., Ethiopia and Uzbekistan) have a record of conflating domestic political opposition elements with Al Qaeda terrorists.
42
and additional set of pressures (not present in relationships with democratic intelligence services, whose masters’ survival is dependent upon the outcomes of elections) that ultimately increases the likelihood of biased intelligence being supplied to the recipient.
Figure 3. The Snowing Effect
Such patterns of behavior are inherently difficult to detect in any transactional intelligence-sharing relationship. This is because intelligence supplied through foreign liaison channels is often stripped of its sourcing information—thus leaving the recipient unable to assess the provenance, reliability and completeness of the intelligence, not to mention the conditions under which it was extracted (Tenet 2004,
21; Scheuer 2007). This puts the recipient in a precarious position: it must pay its supplier for potentially biased intelligence that may ultimately corrupt the judgments of its own intelligence agencies and, in turn, manipulate its government’s foreign
43
policy decisions. To be clear, I am not suggesting that these risks are necessarily unique to transactional relationships involving authoritarian intelligence services.
The consequences of biased (and indeed, false) intelligence and their associated impacts on foreign policy decision-making have also been observed within democratic intelligence sharing relationships, as the ‘Curveball’ example illustrates.39
I am merely suggesting that, to the extent that authoritarian regimes receive a steady flow of rents outside of intelligence channels in exchange for providing certain information, they will be more inclined (relative to democracies) to distort that information in a manner that sustains the cycle.
Political science contributions
It has already been noted that American political science has largely neglected the study of intelligence, whether out of scholars’ concerns for their own professional reputation, personal misgivings about the ethical implications of studying such a controversial topic, or the absence of sufficient and reliable sourcing. There are even
39 Over the course of several years in the early 2000s, Germany’s intelligence service supplied their American counterparts with information concerning Iraq’s alleged weapons of mass destruction (WMD) program. Some of the more incriminating information reportedly came from a since- discredited German informant and Iraqi exile named Rafid Ahmed Alwan, who claimed to have worked during the 1990s for a team responsible for building mobile laboratories to manufacture biological weapons. Information attributed to Rafid, who was known to the U.S. Intelligence Community as ‘Curveball,’ was subsequently cited by the U.S. Government as evidence of Saddam Hussein’s continued pursuit of a WMD capability (CBSNews 2009). That allegation became a primary justification for the 2003 U.S.-led invasion of Iraq. In its 2004 report, the Iraq Survey Group formally discredited Curveball’s allegations. The manner in which Curveball’s allegations were passed to and used by the U.S. Government does not appear to have mirrored the ‘snowing effect’ described above. The German intelligence service seems to have been relatively forthcoming with its American counterparts in expressing concerns about Curveball’s reliability as an intelligence source. Likewise, there is no evidence to suggest that Germany deliberately ‘cherry picked’ information from Curveball and provided it selectively to the United States in order to influence US foreign policy towards Iraq (Drogin and Goetz 2005). The Curveball episode is perhaps better explained as a consequence of poor counter-intelligence exercised by both the U.S. and German intelligence services. I thank an anonymous reviewer from Intelligence and National Security for reminding me of this episode and highlighting its pertinence to my conceptual framework.
44
fewer political scientists who have studied the specific topic of bilateral intelligence cooperation. Security studies specialists are among those select few. They have addressed the importance, challenges, risks and tradeoffs of foreign intelligence relationships in the context of countering transnational terrorism. More recently, a smaller group of IR scholars has begun to apply concepts and principles from international relations theory to the study of this topic.
If the 9/11 attacks elevated terrorism to the top of the security studies research agenda, international intelligence cooperation has not been far behind. Several security studies scholars, pointing to the global reach of transnational terrorist networks and the increased amounts of intelligence required to identify and track them, have argued that the demand for foreign intelligence cooperation has never been greater (Aldrich 2002a; Reveron 2006, 455). Byman goes a step further, suggesting that a country’s “ability to provide intelligence on al-Qaeda” has become a crucial factor in determining the operational value of America’s counter-terrorism alliances (Byman 2006, 769-771). Bensahel makes perhaps the boldest claim, asserting that the international ‘intelligence coalition’ (relative to the financial, law enforcement, military and reconstruction coalitions) is of paramount importance to global counter-terrorism efforts because it is the only coalition that can “provide warning of impending attacks, which can enable countries to take preemptive and preventive actions against those planning such attacks” (Bensahel 2006, 40). A recurrent theme in these studies is the notion that, with regard to bilateral intelligence cooperation, the end generally justifies the means. In other words, cooperation with
‘unsavory’ authoritarian intelligence services (e.g., those of Pakistan, Saudi Arabia
45
and Syria) will remain worthwhile and necessary so long as it yields actionable intelligence against Al Qaeda.
A second group of political scientists has begun to leverage competing branches of IR theory to explain international intelligence cooperation dynamics.
Jennifer Sims and James Igoe Walsh are among the leading scholars in this group.
Sims views the formation of foreign intelligence liaison relationships as one of multiple self-help strategies that a state can employ under conditions of anarchy
(Sims 2006, 196). She argues that the nature and frequency of information exchanges within these relationships will shift in response to the varying structural conditions of the international system. Sims also devises a four-tiered conceptual framework to measure the costs and benefits of intelligence sharing under varying international conditions. In so doing, she strikes a balance between meeting the needs of practitioners (who are faced with managing, justifying and paying for these relationships) and contributing to theoretical development within Intelligence Studies.
There is nevertheless a theoretical tension in Sims’ argument. She appears to move from a neorealist theory of international politics, in which system-level anarchy cultivates balancing behavior as a means of satisfying the state’s ‘drive to survive’
(Waltz 1979), to a rationalist theory of foreign policy that privileges any number of unit-level considerations in a state’s decision to enter, exit, or modify a foreign intelligence relationship. A key question therefore remains unresolved: is international intelligence cooperation properly conceived as a form of balancing behavior (driven by structural, impersonal forces) or as a byproduct of decisions
46
made by rational, utility-maximizing states pursuing their own parochial interests?
An orthodox neo-realist position generally cannot have it both ways.40
Walsh also focuses on the costs and benefits of intelligence sharing, but in contrast to Sims he relaxes the assumption of systemic anarchy. He leverages insights from the relational contracting and transaction cost economics literature to argue that, under certain conditions, states will abandon the anarchy of quid pro quo intelligence cooperation in favor of a hierarchical arrangement. Walsh defines a hierarchical relationship as one in which a dominant state makes all of the key decisions and monitors compliance with agreed-upon rules of sharing. The subordinate state, in return, submits willfully41 to the dominant state and receives a range of benefits (to include foreign assistance) in exchange for its compliance.
Walsh expects that a hierarchical relationship will be formed when a given pair of states has strong and compelling incentives to share intelligence but fears the consequences of defection. The degree of hierarchy is expected to vary in accordance with the fear of defection. Walsh explores two case studies from the early Cold War to illustrate how varying degrees of hierarchy are employed as a hedge against defection (Walsh 2007; Walsh 2010).
By relaxing the assumption of anarchy and presenting a straightforward argument with falsifiable hypotheses and clearly articulated causal mechanisms,
40 Waltz (1979) has argued that any theory that predicts system-level outcomes (e.g., pursuit of a foreign liaison relationship) on the basis of unit-level attributes (e.g., utility-maximizing calculations of states) is ipso facto ‘reductionist’ (47). In outlining a theory that focuses on the system- level, Waltz revivified realism by moving the emphasis from unit-level activity to the effects of the international system. This has been cited as one of neorealism’s chief contributions to IR theory (Keohane 1986, 175). 41 Walsh’s notion of willful hierarchy resembles the ‘client-server’ arrangement described in the practitioner literature, rather than the ‘dictatorial’ arrangement.
47
Walsh has supplied political scientists with an invaluable theoretical baseline for the future analysis of international intelligence cooperation. Having said that, his argument does not (yet) offer a test of competing paradigms. If his stated objective is to explain hierarchical variations in intelligence sharing, then it stands to reason that the competing claims of rival theories should also be considered. Another problem relates to underspecified causal mechanisms, such as the process by which role assignments are made (or taken) in a hierarchical relationship. Walsh’s argument implies that this is a function of intelligence capabilities and know-how, but it is not apparent how this would flow from a relational contracting perspective. Lastly, it remains unclear whether the type of hierarchy that Walsh is describing—a scenario in which two parties vertically integrate their intelligence activities—is a common or infrequent case outcome.42
An analytic framework
The preceding discussion has revealed a dearth of theoretical explanations for bilateral intelligence cooperation. In an effort to identify candidate theories, I have constructed an analytic framework that “organiz[es] the knowledge accumulated to date and provid[e]s abstract representations of that knowledge” (Druckman 2005, 31).
Since the field of Intelligence Studies remains in a pre-paradigmatic phase of theoretical development, this framework will be necessarily crude and incomplete.
This will be an acceptable weakness, however, if the framework succeeds in
42 Richelson (1990), Clough (2004), Lefebvre (2004) and Sims (2006) have each suggested that bilateral cooperation occurs largely under conditions of anarchy, where neither party exerts direct control over the other. They would therefore view a hierarchical arrangement as uncommon and infrequent.
48
identifying variables and relationships (and in turn, established IR theories) that can be leveraged to explain variance in the depth of bilateral intelligence cooperation.
In developing an analytic framework, I emphasize factors related specifically to the dependent variable (bilateral intelligence cooperation) and possible independent variables—that is to say, factors that might conceivably shape the depth of bilateral intelligence cooperation. This information is captured in the conceptual matrix displayed in Appendix I, which contains a sampling of well-known cases of bilateral intelligence cooperation from World War II through the present date. I do not claim that these cases are necessarily representative of the broader universe of bilateral intelligence sharing arrangements. They do, however, cover an extended time period and encompass diverse geographic regions, to include North America,
Western Europe, the former Soviet Union, and the Middle East. It is hoped that a superficial comparison of these cases will shed light on possible causal relationships that merit further exploration.
In the ‘background characteristics’ component of the table, I characterize each case in terms of the countries involved and when cooperation occurred. In the
‘possible drivers’ component of the table, I seek to identify plausible causal variables
(each anchored in an established branch of IR theory) that might account for the outcomes in each case. Several case features that immediately stand out are regime composition, heterogeneity (a type of transaction cost) and structure of the international system. Each of these attributes corresponds to distinct and competing
IR paradigms. Regime type is generally associated with the democratic peace literature, heterogeneity is an essential component of relational contracting, and
49
structure of the international system is a key explanatory variable for neorealism.
Turning to the dependent variable, I considered additional factors that might prove useful in measuring or understanding this concept. Sims (2006) identifies at least two features of interest: relationship symmetry, which captures the extent to which one party derives greater benefits from the relationship relative to the other, and relationship complexity, which refers to the terms of intelligence exchange.
Clough’s (2004) notion of granularity also warrants inclusion, to the extent that it captures the extent of sensitivity (and secrecy) of collaboration in a given intelligence sharing arrangement.
In the next chapter I assess the applicability of two theoretical traditions that are suggested by this framework: dyadic democratic peace theory and relational contracting. From these theories I deduce hypotheses that offer unique and specific predictions about bilateral intelligence cooperation under the scope condition of war against a common enemy.
50
Chapter 3: Theory and Method
In this chapter I examine the theoretical foundations of two IR paradigms – dyadic democratic peace and relational contracting — and illustrate their potential applications to the study of bilateral intelligence cooperation under conditions of war.
I begin by exploring the central arguments of (and counter-arguments against) each perspective, as well as the types of phenomena that they have been leveraged to explain. Next, I consider how each perspective could be applied to the study of bilateral intelligence cooperation. I then formulate hypotheses for each perspective and devise a methodology for testing them against my selected cases. The chapter concludes with a description of procedures that I will use for measuring both the dependent and independent variables.
The dyadic democratic peace argument
If one were to identify a single empirical claim that entertains broad support among IR theorists, it would probably be that liberal democracies do not go to war with one other.43 This so-called ‘dyadic peace’ proposition44 posits that there is something intrinsic to liberal democracies that makes them “less likely to fight
43 One scholar famously called this claim “as close as anything we have to an empirical law in international relations” (Levy 1989, 88). 44 A separate ‘monadic’ argument has received weaker and more qualified empirical support than its dyadic counterpart (Chernoff 2004, 59; Kinsella 2005, 453). There are two distinct claims within the monadic camp. The bold claim asserts that democracies are not likely to engage in conflict with other countries, regardless of their regime type. This position was first advanced in the writings of Schumpeter (1919), but fell out of favor when it became apparent that democracies are vulnerable to involvement in aggressive external behavior, such as war. A more modest monadic claim is that democracies are either slightly less or no more likely than other types of countries to engage in wars. This latter claim has gained limited but weak support in statistical tests conducted by, for example, Rousseau and his colleagues (1996).
51
interstate wars against each other than pairs of states that are not both democratic”
(Ray 2003, 211). The dyadic democratic peace has been corroborated in numerous statistical studies over the years (Babst 1964; Babst 1972; Rummell 1983; Maoz and
Russett 1992; Maoz and Russett 1993; Dixon 1993; Rousseau et al 1996).
Furthermore, as a strictly empirical phenomenon, the absence of war between liberal democracies is widely accepted by even the most outspoken critics of the democratic peace research program (Rosato 2003).45
The fact that democracies do not go to war with one another has compelled some IR scholars to consider whether a regime’s democratic character (or lack thereof) may hold implications for its behavior in other areas, to include alliance formation, trade negotiations, covert intervention, and foreign policy decision-making
(Doyle 1995; Lai and Reiter 2004; James and Mitchell 1995; Young and Urlacher
2007). This dissertation explores whether the dyadic democratic peace argument holds implications for the conduct of democracies that are members of bilateral intelligence alliances. Consistent with the writings of Imre Lakatos (1970), such an extension could form the basis of a progressive research program—specifically by adding new content, “point[ing] the way to new facts and explain[ing] phenomena that the theory was not specifically formulated to explain” (Chernoff 2004, 60).
Before such aspirations can be entertained, however, we must first consider some of the competing explanations of the dyadic democratic peace and determine whether it makes sense to extend and adapt them to the study of bilateral intelligence
45 Some of the most vocal critics have come from within the realist tradition. Early examples include Mearsheimer’s (1990) initial skepticism about the durability and basis of the dyadic peace. Subsequently, realist scholars have cited a number of alternate explanations for the absence of war between democracies, including geopolitical factors, the relative distribution of capabilities and common foreign policy interests (Layne 1994; Thompson 1996; Gartzke 1998; Gowa 1999).
52
cooperation. Arguments traditionally have fallen into one of two camps: structuralist and normative. The following discussion provides a high-level overview of these arguments and considers their applicability to bilateral intelligence cooperation.
Structuralist arguments
According to one popular line of argument, the absence of war between democracies can be explained by the constraining effects of democratic structures.
These structures—which consist of a free press, a domestic political opposition, an electorate, competing branches of government and a large winning coalition, among others—compel democratic decision-makers to proceed with caution as they consider whether to initiate war (Fearon 1994, 577-592; Shultz 1999, 829-844; Siegel 1997, as cited in Shultz; Mesquita, Morrow, Siverson et al 1999, 791-807). Most of these constituencies must be brought on board in advance if the ensuing war is to be viewed as legitimate. Mobilizing these constituencies can be a difficult and time-consuming process (Maoz and Russett 1993, 626). Also weighing on the minds of democratic decision-makers are the perceived odds of military victory. Unlike their authoritarian counterparts, democratic decision-makers face the prospect of being removed from office in the event of a disastrous outcome (Mesquita, Morrow, Siverson et al 1999,
802). In sum, the structuralist view holds that democracies will avoid initiating war unless the public deems the cause legitimate, victory appears imminent and the government is fully invested in victory.
The structuralist view posits that democracies exhibit additional caution in their selection of battlefield opponents. More specifically, they tend to select and fight
53
adversaries that they are at once likely to defeat and strongly committed to defeating.
This suggests that when democracies go to war, they select relatively weak enemies, mobilize more resources and fight harder than other types of regimes. Because democracies are relatively transparent in communicating their intentions and the extent of their capabilities, any two democracies involved in a dispute are unlikely to escalate towards war. Neither can be sufficiently assured of prevailing because each understands the other’s commitment to mobilize for victory and fight hard (Ibid.,
802). Under these circumstances, war becomes a prohibitively expensive option that could cost each side its popular support, and thus its leadership’s hold on power.
Democracies consequently have compelling incentives to negotiate with one other.
By the time that each side has made military preparations for war, so the reasoning goes, the diplomats will have had sufficient time to resolve the conflict peacefully
(Shultz 1999, 236; Maoz and Russett 1993, 626).
Authoritarian decision-makers, by contrast, are less inclined to choose their opponents carefully, fight hard and invest large portions of state resources in military victory. In contrast to democracies, they are more willing to risk war because if the battlefield outcomes are not to their liking, they can simply cut their losses without fear of electoral penalty. This by no means implies that authoritarian leaders are entirely free to do as they please. They too face the risk of being removed from office, particularly if they are unable to satisfy the demands of their small circle of private backers. One implication of this rationale is that authoritarian leaders, when forced to choose between financing an increasingly expensive war and placating their supporters, will favor the latter (Mesquita, Morrow, Siverson et al 1999).
54
Given these attributes of authoritarian and democratic regimes, how does one account for the occurrence of war between them? One possible explanation is that authoritarian regimes view crises in zero-sum terms. In contrast to democracies, which are accustomed to ‘splitting the difference’ on contentious issues, authoritarian regimes may perceive any gain by the other side as a threat to their own survival.
They consequently have an incentive to conceal private information and misrepresent their true preferences to their opponents (Shultz 1999, 236). This absence of transparency creates significant complications for democracies, which can never know for certain what type of adversary they are dealing with (i.e., friend or foe)
(Chan 1997, 53). This in turn can fuel misperception in democracies, whose decision makers are inclined to assume the worst about their authoritarian counterparts, regard negotiations as futile and mobilize the public for war (though it remains unclear why war occurs in some of these instances but not others).
Many of the structural perspective’s underlying assumptions about democratic decision-making structures appear ill-suited to the phenomenon of bilateral intelligence cooperation. This is seen first in the fact that foreign intelligence agencies are instruments of the permanent national bureaucracies that they support. The personnel (intelligence elites) that comprise these agencies are not concerned about losing the next election or alienating members of a fickle winning coalition.46
Intelligence elites also conduct much of their business in secret, which to some extent insulates them from the intense public scrutiny facing elected officials and democratic
46 Nevertheless, democratic intelligence agencies can ill afford to alienate their patrons in the legislative branch, lest restrictions or holds be placed on funding of their activities. As Mark Lowenthal argues, such “control over the budget…is the most fundamental lever of congressional oversight” (Lowenthal 2005, 196).
55
decision-makers.47 The degree of insulation from popular opinion is even greater at the international level, where bilateral intelligence exchange is said to resemble a hidden “world within a world, governed by its own diplomacy and characterized by elaborate agreements, understandings and treaties” (Aldrich 2004, 737). It therefore seems unreasonable to expect that the structural properties of democracies would shape the bilateral conduct of foreign intelligence organizations to the same degree that they influence the foreign policy behavior of democratic decision-makers.
Normative arguments
Normative arguments are rooted in the writings of Immanuel Kant (1795), who argued that universal human aspirations for peace, limited government and freedom of association would provide the impetus for an “ever-widening pacific union” of liberal republics. Kant also cautioned that the march toward perpetual peace would be difficult and uneven. He predicted that liberal republics would feel threatened by closed societies whose leaders did not enjoy the popular consent of their subjects. Such suspicions would, in Kant’s view, provide the spark for violent conflict between liberal and nonliberal states (usually initiated by the former) prior to the onset and consolidation of perpetual peace (Doyle 1986, 1155-1162).48
47 This is not to downplay, deny or in any way underestimate the importance of press exposés, congressional oversight hearings, and Freedom of Information Act (FOIA) lawsuits in informing the general public about intelligence abuses and poor performance. I am merely suggesting that information about intelligence matters is diffused to the public at a lower rate, relative to other, more transparent domains of government activity. 48 On this point, Kant anticipated by over 200 years an argument made by Kenneth Waltz against the dyadic peace argument. Waltz argues that powerful liberal democracies such as the United States are often tempted to ‘save’ oppressed peoples living under authoritarian rule. These ‘wars of national liberation,’ in his opinion, run the risk of overextending American power (Waltz 2000, 13).
56
Drawing from Kant’s insights, a separate group of IR theorists has emphasized the role of democratic decision-making norms in managing domestic- level disputes and regulating political competition. They posit that democratic decision-makers adhere to the implicit norms of non-violent compromise, tolerance and respect for the “rights and continued existence of their opponents” (Russett 1993,
35). These norms are presumed to govern interactions at both the domestic and international levels (Maoz and Russett 1993, 625). This predisposition toward tolerance and peaceful compromise can dissolve, however, when democracies interact with authoritarian regimes. This is because the former harbor deep suspicions about the latter, whose authority is presumed to rest upon violence and oppression (Risse-
Kappen 1995, 37).
Normative explanations of the dyadic peace generally highlight the role of a central causal mechanism: the capacity of democratic regimes to recognize, identify with and trust one another. This democratic regime recognition dynamic has two key dimensions. First, liberal democracies benefit from an inherent presumption of amity when dealing with one another. Second, authoritarian regimes face an inherent presumption of enmity and suspicion when dealing with liberal democracies (Doyle
1986, 1161).
Scholars are of mixed views on how this Janus-faced recognition actually occurs. Most explanations proceed from the assumption that democratic competition at the domestic level is governed by implicit ground rules, or what Schmitter describes as “the contingent consent of politicians acting under conditions of bounded uncertainty” (Schmitter 2002, 184). Contingent consent is analogous to a
57
gentleman’s agreement: each side agrees to respect transactional outcomes that favor the other side, but only on the condition that the ‘winner’ agrees not to exploit its temporary advantage over the ‘loser’ in subsequent interactions. Bounded uncertainty refers to a “predictable and mutually acceptable range” of behaviors that constitute fair play between the actors (Ibid., 185; Dixon 1994, 15-16).
When disputes arise within democracies, participants reach peaceful settlements by adhering to non-coercive norms of conflict resolution. This comes naturally to democratic policy elites, who by virtue of their recruitment and professional training are inclined to favor conciliation over coercion.49 Democratic policy elites know at least two things about each other going into a head-to-head confrontation: neither will resort to violence or coercion and each will be amenable to forging "at least some procedural accommodation" with the other. Taken together, these ground rules and norms constitute a menu of shared understandings that governs intra- and inter-democratic exchange. This creates an atmosphere of mutual trust that provides a resolution to the classic security dilemma. In sum, neither party fears that the other will cheat—and relative gains by one party will not be perceived as threatening to the other’s existence (Ibid.).50
No such menu of shared understandings is presumed to exist in a ‘mixed dyad’, however. This again owes to a perception, commonly harbored by democratic
49 Dixon elaborates: “…[T]he competitive institutional structures of democratic states usually entail a recruitment path limiting such positions to seasoned politicians, career government officials, or experienced elites accustomed to operating in the public realm within the normative standards of bounded competition” (Dixon 1994, 16). 50 Offensive realists see it differently. Given an international competition between two great power aspirants, both can be expected to “tak[e] active measures to gain advantage over their opponent” regardless of its regime type. According to this school of thought, each aspirant will invariably seek to maximize its power as a means of ensuring its own survival (Mearsheimer 2001, 33, 233).
58
policy elites, that authoritarian regimes are “predisposed toward belligerence, since their domestic rule is based on oppression and violence” (Risse-Kappen 1995, 37).
Authoritarian regimes are also thought to favor ‘winner-take-all’ outcomes, which
“deny the loser the power or opportunity to rise again” (Maoz and Russett 1993, 625).
These perceptions compel democracies to behave more selfishly and take a harder line with authoritarian opponents than they would with fellow democracies, although the degree and intensity of confrontation varies from case to case.51 Under these conditions, fears of cheating and concerns about relative gains dominate the actors’ calculations. The security dilemma thus remains alive and well in mixed dyads.
The normative perspective offers a seemingly more plausible fit for bilateral intelligence cooperation. Its insights about democratic policy elites seem particularly well-suited to the decision-making behavior of democratic intelligence elites, who tend to resolve their disputes peacefully through bureaucratic competition. The normative perspective also holds that implicit ground rules and conflict resolution norms govern interactions at both the domestic and international levels. This insight may prove applicable to the behavior of democratic intelligence agencies as well.
The US National Intelligence Estimate (NIE) process, while not illustrative of all types of inter-organizational interactions in the United States Intelligence
Community (USIC), provides a useful example of how implicit ground rules and conflict resolution norms govern behavior among participants in a democratic intelligence community:
51 For one possible explanation of this variation, see Müller and Wolff (2004, 32). The authors contend that a given democracy’s bellicosity towards non-democracies can be explained by its placement along a disposition continuum ranging from militant to pacific.
59
In the United States, national intelligence officers are responsible for preparing estimates. They circulate the terms of reference (TOR) among colleagues and other agencies at the outset of an estimate. The TOR may be the subject of prolonged discussion and negotiation, as various agencies may believe that the basic questions or lines of analysis are not being framed properly. ... Once drafted, the estimate is coordinated with other agencies, that is, the other agencies read it and give comments, not all of which are accepted, because they may be at variance with the drafter's views. Numerous meetings are held to resolve disputes, but the meetings may end with two or more views on some aspects that cannot be reconciled. The DNI chairs a final meeting, which is attended by senior officials from a number of agencies. After the DNI signs the estimate, signifying she is satisfied with it, the estimate becomes hers (Lowenthal 2005, 133-134).
The NIE process represents an example of distinctly democratic rules and norms shaping strategic interaction within the USIC. Implicit ground rules stipulate how the NIE is to be drafted and coordinated. These rules fulfill the role of bounded uncertainty by establishing an acceptable and predictable range of behaviors by the participants. Contingent consent is also on display in the NIE process. Each agency agrees up front to participate, provided that at least two conditions are met: there will be universal adherence among the participants to the aforementioned ground rules, and all will be guaranteed the right to participate in future estimates. Substantive disputes over final language are addressed through a variety of non-coercive conflict resolution mechanisms, to include meetings, footnotes (to accommodate dissenting views) and the DNI’s role as final arbiter.52
52 As one anonymous reviewer reminds me, the NIE process has been “criticized for leading to group think, log-rolling and the watering down of sharp insight.” In light of these criticisms, some might question my use of this example. Here it is worth drawing a distinction between the perceived analytic value of NIEs as a finished intelligence product, on the one hand, and the process by which they are drafted, on the other. My interest lies in the latter. To reiterate: I am not saying that the NIE process necessarily leads to superior analytic products (and by extension, better decision-making), but rather that it provides a useful example of how mutually agreed-upon ground rules and procedures govern behavior among participants in a democratic intelligence community.
60
Can the normative perspective also shed light on interactions between authoritarian and democratic intelligence agencies (i.e., mixed dyads)? To extend the analogy properly, it will be helpful to recall the process by which decisions are made within an authoritarian regime. Competition is typically zero-sum, disputes are resolved through resort to violence and/or coercion, and one individual (the authoritarian leader) has disproportionate influence over final decisions (Young and
Urlacher 2007, 57). By extension, let us assume that similar norms govern domestic- level competition within authoritarian intelligence systems.53 Recognizing the potential threat posed by these powerful ‘states within the state’—and in the absence of norms of bounded uncertainty and contingent consent to regulate their behavior— the authoritarian leadership has an incentive to play the various security and intelligence organs against each another. As Knight argues, this is more or less what happened in the Soviet Union, where the Soviet intelligence and security services reportedly engaged in unregulated and often adversarial competition with one another
(Knight 1988, 85-86).54
What might these normative properties of authoritarian and democratic intelligence services tell us about the nature of their interactions? One implication seems clear: they are likely to exhibit high levels of mutual distrust.55 Unable to
53 The plausibility of such an assumption is supported by the experience of the former Soviet Union, whose secret police and intelligence services were “central to the structure of the one-party state and to the systems of repression and social control which [sought] to suppress all challenges to its authority” (Andrew 2004b, 177). 54 To cite one example, GRU Colonel and western intelligence asset Oleg Penkovsky reportedly informed his handlers that the GRU and KGB “constantly compete[d] with each other in espionage” during the early 1960s and regarded each other as ‘the enemy’ (Ermarth 1965). 55 Of course, some level of mutual distrust is arguably present in all bilateral intelligence transactions, regardless of the regime types involved. My contention is merely that there is a substantial difference in degree of suspicion between democratic services, on the one hand, and
61
assess confidently the motives of the other side—and in the absence of a menu of shared understandings—each can be expected to gravitate toward strategies that ensure absolute and immediate gains for both sides or no gains for either.
This loss-averse and transactional approach to intelligence cooperation, widely known among intelligence practitioners as ‘quid pro quo’, is predicated on the assumption that one side will not share anything of value unless it is assured of receiving something of greater or equal value in return (Clough 2004, 601; Lefebvre
2003, 537). These concerns become more salient in an atmosphere of mutual distrust—particularly when one considers the sensitive nature of intelligence goods, the significant costs of acquiring and protecting them, and the potential risks to national security if they get into the wrong hands (Reveron 2006, 467-468). In such an atmosphere, it seems likely that neither party will be inclined to share intelligence
(or cooperate in intelligence matters) ‘on credit.’
Figure 4 suggests several potential implications of the normative perspective for bilateral intelligence cooperation. It illustrates a hypothetical one-off scenario involving two foreign intelligence organizations from different countries (A and B) that share a common adversary. A asks B to share an unspecified intelligence good concerning this adversary.56 Given this set of circumstances, what cooperative strategies are available to B? Drawing upon insights from the normative dyadic peace argument, I suggest that the answer may well depend on what type of intelligence dyad is involved. democratic and authoritarian services, on the other. 56 In making this request, A assumes—or has reason to believe—that B possesses the requested intelligence good. Whether B actually possesses this good is another matter.
62
Figure 4. Bilateral Intelligence Cooperation Heuristic
Mixed intelligence dyads are presumed to take a relatively narrow view of what constitutes acceptable terms of intelligence cooperation. In the hypothetical one-off example, I submit that B is likely to insist upon receiving something specific in exchange for its cooperation (quid pro quo) or reject the request outright
(categorical defection). Accordingly, B will be unwilling to supply x1 ‘on credit’ or with ‘no strings attached’ to A. This will be the case regardless of which regime type in the mixed dyad is playing the role of B. While it is possible that either party (or both parties) might be willing to consider the possibility of extending this game into the future, neither will give much thought to the implications of the one-off outcome for the ‘shadow of the future’. All things being equal, mixed dyads will tend to focus on the transaction at hand, rather than the prospect of future transactions.
Fully democratic intelligence dyads, by contrast, are presumed to take a more flexible and expansive view of what constitutes acceptable terms of intelligence
63
cooperation. Consequently, there is a greater likelihood that B will gravitate towards cooperative strategies that ‘keep the game going’ to one another’s mutual benefit. To the extent that the shadow of the future enters B’s calculations, B will be more inclined to consider strategies that do not necessarily guarantee it direct and immediate payoffs. The dyadic democratic peace argument provides a possible explanation of why this might be the case. The aforementioned menu of shared understandings (i.e., bounded uncertainty and contingent consent) will help to dampen B’s suspicions about A’s motives. B may consequently be willing to consider cooperative strategies beyond an immediate quid pro quo, such as a counter-offer or conditional cooperation.
Hypothesis
Democratic peace arguments are often leveraged to explain dichotomous
‘snapshot-in-time’ phenomena, such as the outbreak of war or the formation of alliances. I depart from this approach to trace the ebb and flow of intelligence cooperation over time between different pairings of regime types under common background conditions. I expect that an examination of bilateral intelligence cooperation under conditions of war will enable me to test for the presence of contingent consent and bounded uncertainty, which in my view are the principal signatures of the regime recognition dynamic.
Framed in this context, my research question becomes: to what extent does the democratic composition of a wartime cooperative intelligence dyad (WCID) shape the depth of bilateral cooperation between the two members during the course of said war? In posing this question I make the following case assumptions:
64
• Bilateral intelligence cooperation has already commenced. This assumption enables me to focus on the dynamics of intelligence exchange after a liaison relationship has been established, rather than on the factors that may have led to the formation of the relationship in the first place.
• While there may be a large number of conceivable WCID regime permutations, each of them falls somewhere within a bounded property space. The property space can be conceptualized as an x-y graph whose axes mirror the Polity Project’s ‘polity2’ variable, which assigns a single score between -10 (autocracy) and 10 (democracy) for a regime (Marshall, Jaggers et al 2011). This property space is illustrated in Figure 5. This assumption enables me to define the range of potential WCID combinations and therefore place boundaries upon the universe of cases.57
• Each foreign intelligence organization is by definition subordinate to a single national government. A given foreign intelligence agency will therefore carry out functions consistent with the expressed will of its government. This assumption allows me to treat foreign intelligence organizations as extensions of the governments that they serve, thus excluding (for purposes of this argument) the possibility of ‘rogue agencies’ operating under their own institutional prerogatives.58
• Foreign intelligence organizations are ‘hard-wired’ to proceed with caution and size up their counterparts during the early stages of bilateral intelligence cooperation.59 By necessity, this period will be characterized by mutual suspicion and a certain amount of conflict and misunderstanding—regardless of the regime types involved. For instance, one state may accuse the other of ‘free riding’ off its intelligence (i.e., getting something for relatively nothing) or suspect the other of sharing privileged information with third parties.
From these assumptions I deduce the following hypothesis:
57 Despite their appearance in the WCID property space, I am consciously excluding ‘fully autocratic’ WCIDs from my analysis. The question of why autocratic nations might engage in varying degrees of intelligence cooperation, though interesting, falls outside the scope of my current research aims, not to mention the claims of dyadic democratic peace theory. 58 It is important to note that this assumption does not always hold in reality. For example, some regimes operate as ‘independent security states’ in which intelligence services are so autonomous and removed from the day-to-day operations of government that they are effectively answerable to no one (Keller 1989; Boraz and Bruneau 2006, 38). 59 According to Sims, most experienced intelligence services engage in preliminary counterintelligence activities against their foreign partners. This tendency to ‘run the traps’ is a useful means of determining whether unknown third parties are benefiting from an intelligence relationship that is ostensibly bilateral (Sims 2006, 202).
65
H1: The democratic composition of a WCID is expected to shape the depth of bilateral governmental intelligence cooperation observed between the two parties. More specifically, a fully democratic WCID is expected to reach and maintain higher levels of cooperation during the war than a WCID consisting of a democracy and an autocracy.
Figure 5. Regime Type Property Space for WCIDs
In sum, I expect to see in each case an initial period of ‘sizing up,’ regardless of the WCID’s regime composition. This initial period will be characterized by a certain degree of mutual suspicion and limited intelligence cooperation. In fully democratic WCIDs, this period should eventually give way to an increase in the depth of observed intelligence cooperation as the recognition dynamic takes hold. By contrast, when the WCID is composed of a democracy and an autocracy, I expect
‘sizing up’ to persist indefinitely throughout the war with a negative long-term effect on the depth of observed intelligence cooperation. In both cases I expect to see evidence of the purported causal mechanism at work. Fully democratic dyads, operating under the shared democratic norms of contingent consent and bounded
66
uncertainty, will work through their difficulties in the interest of ‘keeping the game going’ to one another’s mutual benefit. Mixed dyads, on the other hand, will operate under conditions of unbounded uncertainty and the absence of contingent consent.
Unable to anticipate the expected range of their authoritarian partners’ behaviors and unwilling to foreswear the use of coercive tactics against them, democratic members in such relationships are expected to become increasingly preoccupied with relative gains and the risks of partner defection.
The relational contracting argument
An alternative explanation posits that states enter into binding cooperative arrangements when they are no longer able to obtain sufficient amounts of national security at an acceptable price through their default strategies of self-help. This line of argument, inspired by the transaction cost school of economics popularized by
Coase (1937) and Williamson (1979), views cooperative arrangements as analogous to firms. In other words, alliances will emerge in response to rising transaction costs faced by unitary actors (states) seeking to survive. Another term for this is relational contracting. IR scholars have subsequently invoked relational contracting arguments to address the creation, persistence, and rational design of international institutions.60
Collectively, thinking about international cooperation as a contracting problem has helped to move the field beyond the question of why cooperation occurs in the first place to a more complex puzzle: why states choose institutional arrangements that vary in terms of their bindingness and structural commitments.
60 A comprehensive review of this scholarship is not within the scope of this study. Each of the aforementioned questions is treated in much greater depth by Keohane (1984); Abbott and Snidal (1998); and Koremenos, Lipson and Snidal (2001). Gilligan (2007) also offers a comprehensive and critical assessment of relational contracting approaches.
67
After providing a brief overview of the relational contracting perspective and its origins within the economics literature, the following section examines the causal claims of two complementary explanations within the relational contracting literature and then considers their potential applications to the study of bilateral intelligence cooperation.
Origins within the economics literature
A common starting point for most relational contracting arguments is the notion of rational, independent and utility-maximizing economic actors engaged in arms-length transactions. In this default mode of economic organization, buyers pay no heed to the identity of the supplier. They seek only the optimal price for a particular good or service. In select instances, however, a specific buyer and seller may see fit to contract with one another exclusively over an extended time period, such that the price mechanism is superseded in favor of maintaining the relationship
“for its own sake” (Coase 1937, 390).
These types of arrangements can prove difficult to sustain in the long run due to the zero-sum self-interests of buyer and supplier. Changing circumstances inevitably present new opportunities for each party to maximize its individual gains through haggling and the threat of defection (Williamson 1979, 242). Failure to curb this sort of behavior can lead to undesirable results for both parties, to include forfeiture of the asset-specific investments and unique economic benefits upon which the relationship was forged.
Relational contracting provides a mechanism for attenuating this sort of opportunism. Should both parties determine that their relationship is worth
68
preserving, they may choose to exit the market and form a more binding relationship.
This generally occurs when buyer and supplier decide that their transactions yield sufficient benefits to warrant the abandonment of the default organizational mode
(classical contracting) in favor of dedicated long-term cooperation (relational contracting).
Williamson refers to these benefits as “idiosyncratic economies,” which accrue exclusively to participants in transaction-specific exchange. Idiosyncratic economies consist of communication economies (e.g., specialized language, standard operating procedures, shared knowledge, and trust) as well as cost economies (e.g., lower production costs enabled by special purpose production equipment) that do not and cannot exist outside of the special relationship. The only way these benefits can be maintained is through the assurance of a continuing relationship. Idiosyncratic economies will vanish when and if the transactional relationship is dissolved, since the assets (which by definition represent sunken costs) cannot be reinvested or redeployed to other activities (Williamson 1979, 241).
Within the economics literature, relational contracting arrangements typically assume the form of a bilateral or a unified governance structure. Bilateral structures preserve the autonomy of both buyer and seller and are premised on the notion that the complete range of transactional contingencies cannot possibly be anticipated in a single contract. Both parties thus reconcile themselves to negotiating any future modifications through follow-on agreements. Typically these modifications will focus on the less contentious issues of quantity and delivery, as opposed to price. The goal of these deliberately incremental changes is to maintain the relationship by
69
“restricting adjustments to those where the hazards [of defection] are least”. Unified governance structures, by contrast, merge the buyer and seller into a single, vertically integrated organization. Integration eliminates concerns of opportunism and trust, since all decisions concerning quantity, delivery, and pricing are made internally by an overarching authority (Ibid., 251-253).61
Most relational contracting arrangements highlight the role of transaction costs in shaping the formality and bindingness of a cooperative arrangement. Those costs, in turn, are decomposed into several elements. The first and most important of these components is typically environmental uncertainty. The shadow of the future and unforeseen dangers compel the parties to craft new arrangements to address potential contractual gaps that could arise in the future. The second component is transactional frequency. When buyer and supplier engage in recurrent transactions
(or have reason to suspect that they will so engage), both parties are inclined to create relational contracting structures to govern exchange. Such structures are the only conceivable means by which they can hope to recoup the initial set-up costs of their cooperative arrangement. The third component is asset specificity, which Williamson defines as transaction-specific investments that cannot conceivably be removed and redirected to other economic activities. As asset specificity increases, so too does the incentive to engage in relational contracting (Ibid., 240-254).
Williamson’s writings eventually found resonance within the field of
International Relations, where scholars began leveraging its insights to explain why states would willfully surrender their sovereignty and enter into binding cooperative
61 Unilateral governance structures are outside the scope of this study, which focuses on bilateral intelligence cooperation between sovereign state actors (which by definition have not ceded their sovereignty to an overarching institution or authority).
70
arrangements with one another (Keohane 1982). In relational contracting, IR scholars found ready analogies with economic actors (states) and the unique forms of cooperation (institutions) that they occasionally create in lieu of adhering to the dictates of the price mechanism (anarchy). In accordance with this reasoning, it was often posited that international institutions should arise in circumstances where “the relative transaction costs of creating them, amortized over the expected lifetime of the regime, [were] ..sufficiently small” (Gilligan 2007, 8).
Weber’s anarchic argument
Within the IR relational contracting literature, one popular line of argument posits that the degree of bindingness62 observed in cooperative security arrangements is shaped by the level of external threat and transaction costs facing the participants, which typically are sovereign political actors (i.e., states) acting in an anarchic, self- help environment. When the level of external threat and transaction costs is low, the participants are expected to pursue relatively informal and non-binding alliances. By contrast, when these levels are high, the participants are expected to enter into more binding arrangements such as formal alliances and confederations (Weber 1997).
Such arguments are exemplified in the writings of Weber (1997, 2000).
Despite reserving many of the key insights from relational contracting, Weber makes several notable departures from the transaction cost economists. First, she discounts the causal effects of transactional frequency in this particular context, arguing that
62 Weber defines bindingness as “curtailment of sovereignty in exchange for greater institutionalization” and a “measure [of] the degree of structural commitment [that] allies agree to make” (Weber 1997, 322; 324). Later in this chapter I will offer a slightly different definition of this term which highlights the central role of structural commitment but de-emphasizes the role of sovereignty in favor of the clarity of the parties’ commitment to one another.
71
threatened states (in contrast to economic actors in a marketplace) are much more concerned with survival than they are with recouping the set-up costs of a cooperative security arrangement. Second, she operationalizes her remaining independent variables somewhat differently. Specifically:
• The level of external threat is expected to vary in accordance with the states’ perceptions of a potential aggressor’s military capacity and the geographic proximity of the adversary to said states.
• Transaction costs further shape the bindingness of cooperation through the observable effects of:: . Structural uncertainty, measured in terms of the number of Great Powers in the international system and the presence or absence of clear, consistent signaling between states. . Asset specificity, characterized as “highly trained men, specialized equipment, or specific sites that states otherwise would not invest in or could not readily shift to.” . State heterogeneity, seen in the degree to which factors such as “language problems, misunderstandings, and disagreements” must be resolved through “translation, arbitration, and costly coordination” (Weber 1997, 333).
A key strength of Weber’s adaptation of the relational contracting perspective is that it defines a specific set of conditions under which states feel compelled to abandon exclusively self-help strategies in favor of more binding security arrangements. With subtle modifications, this argument could potentially be extended even further as an explanation of bilateral intelligence outcomes.
Nevertheless, a potential weakness of Weber’s argument is seen in its implicit assumption that states facing high levels of transaction costs and external threat have only one option available to them: develop more binding anarchic arrangements
(e.g., alliances or confederations) or succumb to defection (Ibid., 323). This does not allow for the possibility of hierarchical modes of cooperation in which one state assumes direct control over another’s security. The importance and implications of
72
the hierarchical mode of cooperation are discussed in greater detail in the next section.
Walsh’s willful hierarchy argument
Economists and IR scholars have long envisioned scenarios in which states might surrender their sovereignty to some external, non-state entity (e.g., in an integrated governance structure such as a confederation) or enter into formal cooperation on equal terms with each other (e.g., in a bilateral governance structure).
Less consideration has been given to the possibility of one state voluntarily surrendering some of its sovereignty to another. Such a scenario was long considered anathema to IR theorists who subscribed to the neorealist assumption of
Westphalian sovereignty, in which states engage each other as equals amidst the backdrop of international anarchy (Waltz 1979).63
Walsh was among the first to challenge this assumption in his hierarchical account of relational contracting, wherein nominally sovereign intelligence organizations enter into binding cooperation with one another voluntarily on less- than-equal terms (Walsh 2007; Walsh 2010). Walsh’s argument complements and extends upon Weber’s anarchic argument in several important respects. In addition to highlighting a third possible type of binding arrangement between intelligence organizations, his argument lends itself well to the generation of observable
63 It was arguably not until the late 1990s that the movement to ‘bring hierarchy back in’ began to develop momentum. At the forefront of this movement was Lake, who articulated a number of arrangements (sphere of influence, protectorate, informal empire and empire) in which states conceivably could enter into security relationships on less-than-equal terms (Lake and Powell 1999; Lake 2003).
73
implications that help to explain why some binding arrangements assume the particular forms that they do.
Walsh accomplishes this by picking up where Weber left off. Assuming that the essential background conditions of Weber’s anarchic argument (viz., high levels of external threat, uncertainty and transaction costs) have already been satisfied, he seeks to identify additional factors that influence whether the binding arrangement is anarchic or hierarchical in nature (Walsh 2010, 99). An example of an anarchically binding arrangement would be one in which formal agreements or institutions are created to promote greater efficiencies in information exchange, but no constraints are imposed on either party’s decision-making authority (Ibid., 2010, 25).64 An example of a hierarchically binding arrangement would be one in which a state assumes direct control of another state’s intelligence activities as a means of resolving the problem of defection (Ibid., 99-104).65
Walsh contends that hierarchical arrangements are most likely to emerge when both parties expect to derive significant benefits from intelligence cooperation, at least party perceives a high risk of defection by the other, and there is a clear dominant and subordinate party. The dominant party generally possesses a greater share of material resources and is presumed to have more options at its disposal (i.e., alternatives to cooperation) for intelligence gathering against the common enemy. It is ultimately the dominant party’s ability and willingness to monitor and enforce
64 Such arrangements are similar in form to the bilateral governance structures described in the transaction costs economics literature, in which market actors create institutions to address the less contentious issues of quantity and delivery, rather than price. 65 This offers a distinct departure from the economics literature, in which bilateral governance structures preserve the autonomy of both parties and their status as relative equals.
74
compliance with formally negotiated rules which allows hierarchical intelligence cooperation to thrive and persist.
The dominant party cannot simply will a hierarchical arrangement into being, however. The weaker party must also agree to the proposed terms of cooperation.
Walsh contends that this is most likely to occur when both parties “stand to gain from sharing, have strong concerns about their partners’ reliability, and conclude that the costs of creating and maintaining the hierarchy are smaller than the joint gains that they could expect” (Ibid., 59-60). For the weaker state, these gains often consist of some tangible risk premium (i.e., rents) that it stands to receive in exchange for its compliance (Ibid., 17-25).
I believe that a hybrid relational contracting argument, built upon both
Weber’s anarchic and Walsh’s hierarchical arguments, has the potential to explain why and how different forms of bilateral intelligence cooperation emerge under highly stressful background conditions. In the next section I attempt to construct such an argument by formulating some entry-level assumptions and a testable hypothesis.
Hypothesis
To what extent can a synthesis of the anarchic and hierarchical variants of the relational contracting perspective shed light on the dynamics of bilateral intelligence cooperation? I proceed by making some entry-level assumptions:
• The WCID has already been formed and its members (foreign intelligence services) act as extensions of the governments that they serve.
• WCID members are inherently pre-disposed to dealing cautiously with each other during the early stages of cooperation.
• The decision to engage in (or withdraw from) bilateral intelligence cooperation is an iterative choice that can occur potentially numerous
75
times during the observation period (in this case, the duration of a war).
• Any bilateral cooperative arrangement involving the exchange of intelligence inherently entails a certain degree of asset specificity.66 Since intelligence by definition involves secrecy, a certain portion of its value (whether as information, tradecraft, or knowledge of collection capabilities) becomes irretrievable once one party has shared it with another. An important implication of this assumption is that all cases of intelligence cooperation will entail a certain base level of asset specificity which naturally accumulates over time, so long as cooperation continues.67
• The level of external threat will be necessarily high in all WCIDs examined in this study, given that they are aligned against common wartime enemies. This assumption enables me to hold constant the variable of external threat and explore the causal effects of transaction costs.
From these assumptions I deduce a relational contracting hypothesis and a corresponding set of contingent predictions:
H2: The bindingness of a bilateral intelligence relationship (WCID) will vary in accordance with the level of external threat and transaction costs facing its members. More specifically, relatively greater degrees of bindingness will be observed in WCIDs facing significant external threats and high transaction costs. Conversely, relatively lower degrees of bindingness will be observed in WCIDs facing weak external threats and low transaction costs.
It is important to acknowledge specifically what is implied by this hypothesis.
Bilateral intelligence cooperation is expected to become more binding between partners facing very high levels of external threat and significant transaction costs.
The opposite outcome (informal and ad hoc exchange) is expected for WCIDs facing low levels of threat and relatively minor transaction costs. A probabilistic claim is
66 Walsh (2007) argues somewhat differently, asserting that the only intelligence assets that meet this definition are those—such as networks of human agents and listening posts—which focus on highly specific problems and are incapable of being transferred to other targets or intelligence problems (157-158). 67 Historian Bradley Smith argues that asset specificity played an important role in the continuation of UK-US intelligence cooperation after World War II: “…[O]nce the highly complex process of secret-information gathering used by Britain and the United States had been shared between them, their secrecy could be maintained only if the partnership was maintained” (Smith 1993, 157).
76
also implied in this hypothesis: given two WCIDs aligned against the same enemy during the same war, cooperation will be more binding in that WCID which exhibits relatively higher values over the independent variables (external threat and transaction costs) during the course of the war.
Methodology
Research objectives
As noted in Chapter Two, the academic study of international intelligence cooperation has long suffered from under-theorization and a dearth of empirical data.
Addressing these gaps in the literature requires a more flexible dissertation structure than the somewhat rigidly defined 'ideal types' that are frequently utilized in the political science literature (Van Evera 1997; George and Bennett 2005). When existing theories are extended to novel problem domains and leveraged to explain different types of outcomes, new variables must often be defined and examined. Such an approach is consistent with a ‘theory development’ dissertation design, which leverages elements of both induction and deduction to produce a modified theory that can be extended against a new problem domain (George and Bennett 2005, 114).
Once a modified theory has been constructed, its causal claims cannot be accepted at face value. Ideally, they should be tested—if even on a preliminary basis—against real-world data and evaluated. This is more in line with a ‘theory testing’ approach68 to theory building, in which deliberately selected case studies are employed to
68 Theory testing approaches generally entail at least four tasks: (1) identifying applicable and competing theoretical explanations about a given class of phenomena; (2) selecting historical cases about which said theories make strong and unique predictions; (3) ascertaining the extent to which each theory’s predictions ‘fit’ with the case outcomes; and (4) determining whether any tentative typological inferences can be drawn (George and Bennett 2005, 117).
77
“strengthen or reduce support for a theory, narrow or extend the scope conditions of a theory, or determine which of two or more theories best explains a case, type, or general phenomenon” (Ibid., 75).
One potential criticism of this hybrid approach is that the two theories under consideration are not directly comparable because they purport to explain different features of the dependent variable. Whereas the dyadic democratic peace argument makes predictions about the relative degree and extent of bilateral intelligence cooperation over time, the relational contracting argument makes predictions about the structure and form of cooperative arrangements. Such a criticism would be problematic if the central objective of the dissertation were to devise and test rival hypotheses that make competing (read: mutually exclusive) claims about the depth of bilateral intelligence cooperation. As noted earlier, however, such a research objective is analytically premature because bilateral intelligence cooperation is only beginning to receive attention as a study variable within IR. A certain amount of theoretical development must still be done to identify suitable theoretical traditions, extend their arguments properly, and construct preliminary tests (we can call them plausibility probes) to assess their potential application to the larger universe of cases.
This hybrid dissertation design is also vulnerable to criticisms that are leveled against theory development and theory testing designs, respectively. Perhaps the most commonly acknowledged risk of a theory-development design is confirmation bias, which arises when a theory is tested against the same case from which it was derived. In cases when this risk cannot be mitigated through the identification of a separate and independent case, one can leverage the theory’s ability to explain and
78
account for other evidence within the original case. Theory-testing approaches, for their part, are sometimes prone to the emergence of a ‘strawman’ argument in which a particular theory is prematurely dismissed because it has been forced to explain outcomes that fall outside of its temporal or topical scope. One can guard against the strawman scenario by paying careful attention to the “range of institutional settings, cultural contexts, time periods, geographic settings, and situational settings” to which a particular theoretical explanation’s findings apply (Ibid., 119).
To the extent that they make use of small-n case studies, theory-development and theory-testing dissertations also incur the risks of confounding and omitted variable bias. These risks can be mitigated through deliberate and purposive case selection—in other words, identifying “instances where explanations make unique predictions about the process or outcome of the case” (Ibid., 118). As Sagan (1993),
Khong (1992) and others have illustrated, an effective case selection strategy draws into relief the competing causal claims arguments and case-specific predictions of rival theories.
Case selection criteria
Intelligence, as previously discussed in Chapter One, is an expansive term whose core attributes are thought to consist of secrecy, information, processes, and activities. Each of these attributes is present to some degree in the various intelligence disciplines or ‘INTs,’ most notably human intelligence (HUMINT), imagery intelligence (IMINT) and signals intelligence (SIGINT). Intelligence functions are also performed by a broad range of organizational actors, to include intelligence services, military service elements and branches of the diplomatic corps.
79
This multiplicity of attributes, disciplines and actors presents difficulties for case selection. So too does the dearth of available primary source information. In an effort to address these obstacles, I developed the following case selection criteria:
Strive to hold constant as many competing independent variables as possible. George and Bennett (2005) argue that “confident estimates of causal effects…are possible in case studies only when there is a very well-controlled before- after case comparison in which only one independent variable changes” (25). This is a difficult condition to satisfy in the data-impoverished field of Intelligence Studies.
Nevertheless, the criterion is worth pursuing for at least two reasons. First, it can aid in the identification of appropriate scope conditions for future intelligence theorizing.
Second, it can minimize the number of ‘alternate causality’ arguments to be considered in the future.
Incorporate war as a scope condition. When two countries face the prospect or reality of war against a common and mortal foe, their incentives to share intelligence against that foe increase dramatically, since each party’s survival conceivably lies in the balance. As an environmental stressor, war should produce systematic spikes in the observed ‘amplitude’69 of bilateral intelligence cooperation against the common foe during the observation period. This in turn should facilitate measurement of the dependent variable across cases.
My decision to incorporate war as a scope condition was predicated on several additional assumptions. First, I assumed that bilateral intelligence cooperation is conducted primarily through conventional channels during peacetime (e.g.,
69 I am using the term as it is commonly defined in physics: “the absolute value of the maximum displacement from a zero value during one period of an oscillation” (Random House, 2008).
80
established networks of foreign liaison officers), when the demand for intelligence- sharing is relatively low. In the absence of a systematic environmental stressor such as war, I expected that it would be difficult to detect fluctuations in bilateral intelligence cooperation and conduct cross-case temporal comparisons. I consequently decided against selecting cases of intelligence cooperation that occur during peacetime. Second, I assumed that bilateral intelligence cooperation is inherently more salient (and therefore more observable) during war, when the demand for intelligence-sharing increases and additional elements of the national government apparatus become involved in these efforts out of strategic and operational necessity.
Control for realism. In the absence of a prevailing explanatory paradigm, many intelligence scholars have anchored their arguments within ambiguously defined ‘realist’ frameworks (Fry and Hochstein 1993, 17).70 These arguments generally proceed from the functionalist premise that intelligence-related outcomes can be explained as a consequence of states acting in their own national interests, and/or in response to the pressures of the international system.71 Proponents of these types of arguments might object to a research design that does not explicitly develop and test a rival realist hypothesis. In anticipation of such objections, I offer two observations. First, this study incorporates a phenomenon (bilateral intelligence cooperation) and unit of analysis (bilateral intelligence alliances) that are incompatible with most realist frameworks, which tend to focus on great power
70 Scott and Jackson have picked up on this as well: “It is interesting to note that, while there exists an implicit (and sometimes explicit) assumption that the study of intelligence falls within the realist camp, contemporary neo-realist writers have largely ignored intelligence in their reflections (Scott and Jackson 2004, 147). 71 This is seen in the practitioner and security studies literature on foreign intelligence liaison. See, for example, the works of Richelson (1990), Lefebvre (2003), Clough (2004), Lander (2004), Reveron (2006), Byman (2006), and Bensahel (2006).
81
balancing behavior among states.72 Second, the cases in this study effectively control for realism because they occur under conditions (namely, war) in which realist outcomes and background conditions are already present. In other words, war has already begun and intelligence alliances have been formed against common foes.
Select cases in which the units of analysis perform similar intelligence functions. In the pre-paradigmatic field of Intelligence Studies, it remains unclear whether all intelligence activities can be logically grouped together. Certain dynamics present in bilateral military intelligence cooperation, for example, might be inapplicable to or absent from bilateral intelligence cooperation between law enforcement agencies. It therefore seemed prudent to select cases in which the units of analysis are roughly equivalent in terms of their intelligence functions.
Select cases exhibiting extreme variance over the dependent variable.
Geddes (1990) argues that selecting upon a specific, single value of the dependent variable yields an unduly restrictive universe of cases, which in turn can lead to the generation of erroneous inferences about the broader population. She argues that this problem can be addressed by identifying the relevant universe of cases pertinent to a research problem and then randomly selecting cases from within that universe.
Nevertheless, as Bennett and Elman (2006) observe, the boundaries of the universe of relevant cases are often unknown. Under these conditions it becomes impossible to identify a set of cases that fulfills the statistical ideal of ‘sufficient variation’ on the dependent variable. When that occurs, a suitable (though admittedly imperfect) backup strategy is to select cases lying on opposite ends of the (known) dependent
72 While neo-realists have developed arguments to explain patterns of alliance formation more generally, they generally have not considered the question of why varying levels of cooperation occur within these alliances (Walt 1987; Snyder 1990).
82
variable continuum. Given the relative dearth of case data on my chosen topic, I sought to select cases in which the observed depth of intelligence cooperation was either relatively high or relatively low. In doing this, I recognized that subsequent empirical measurement could reveal that the expected values were incongruous with the actual case outcomes.
* * * * * * * * * * *
Working from the above criteria, I selected two instances of bilateral military intelligence cooperation from World War II: the Anglo-Soviet and Anglo-American
WCIDs. Both come reasonably close to satisfying the aforementioned case selection criteria:
• Hold constant competing independent variables. This condition is partially satisfied in the sense that both cases occur during the same time period, amidst the same set of external threat conditions and against a common set of enemies. There are admittedly several conditions that are not held constant across both cases, such as primary spoken language and national economic strength. I will consequently be unable to rule out the potential causal effects of these factors, regardless of whether congruence testing yields results that are consistent with either of the rival hypotheses.
• Incorporate war as a scope condition. This condition is satisfied for both cases, which occur during a war that clearly threatened the future survival of all Allied participants.
• ‘Control’ for realism. Each case controls for the effects of realism by focusing on the period immediately following the first reported wartime military intelligence liaison overture. For the Anglo-American WCID, this period probably commences in the summer of 1940, when Colonel William Donovan traveled to London on a special liaison visit authorized by President Roosevelt. By this point Germany and the UK were already at war and the US, despite its official neutrality, was arguably beginning to appreciate the gravity of the Axis threat.73 In the case of the Anglo-Soviet WCID, cooperation appears to commence sometime in mid-1941, just as the British and Russian militaries were establishing intelligence missions
73 Just prior to Donovan’s visit, the British Army had made its initial pitch for intelligence cooperation with the Americans against Axis communications (Gladwin 1999, 119).
83
in Moscow and London, following the initiation of Operation Barbarossa (Smith 1996, 33).
• Analogous intelligence functions. Both cases focus on military-to- military channels of intelligence cooperation.
• Variance on the dependent variable. With respect to the dyadic democratic peace argument, both cases appear to fall on opposite sides of the bilateral intelligence cooperation intensity continuum. Historians generally contend that the Anglo-American WCID evolved over time into a high-intensity cooperative relationship, whereas the Anglo-Soviet WCID—despite getting off to a promising start—lingered primarily on the lower end of the continuum for the duration of the war, whereupon it ultimately dissolved. A similar variance is seen in the dependent variable of the relational contracting argument for both cases. Specifically, the Anglo-Soviet WCID was not governed by the sorts of binding agreements that were negotiated between the UK and US during the course of the war. Of course, these preliminary assessments are no substitute for empirical measurement, which will occur in the next two chapters.
Data availability
An important question to consider, given the inherently secretive nature of this topic, is whether sufficient and reliable primary source information is available to support systematic scholarly inquiry. I believe there is. As previously discussed in
Chapter One, the quality and availability of intelligence data writ large has improved
(albeit in fits and starts) since the end of the Cold War. Both the US and UK
Governments have continued to declassify materials concerning cooperative military intelligence activities directed against the Axis powers from at least 1939 through
1945.
Vast stores of these materials are available at the US National Archives
(Archives II) in College Park and the UK National Archives in Kew, England.
Archives II houses at least three collections of interest to this study: The Historic
Cryptographic Collection (Record Group 457), Records of the War Department
84
General and Special Staffs (Record Group 165), and Records of the U.S. Joint Chiefs of Staff (Record Group 218). The UK National Archives, for its part, houses a complementary collection of records that touches upon various aspects of World War
II-era military intelligence liaison with British allies. The records pertaining to the
Government Code and Cypher School (HW 14 and HW 50), the war diaries of the
British Military Mission to Moscow (WO 178/25, WO 178/26 and WO 178/27) and the off-the-record monthly notes of the Signals Intelligence Liaison Unit (SLU) in
Washington (HW 57/5) proved to be of particular relevance in this regard. Both collections allow for a thorough examination of Anglo-American and Anglo-Soviet wartime cooperation from a British perspective.
Testing and measurement
Testing procedures
Given my research objectives, I felt that it was appropriate to employ a combination of the congruence method and observable implication testing. George and Bennett define the congruence method as a technique in which the investigator assesses the ability of a theory to explain and/or predict a case-specific outcome. In order for this to work, the theory must “posit a relation between variance in the independent variable and variance in the dependent variable” (George and Bennett
2005, 181). While the congruence method cannot be used to determine whether a posited causal relationship in a given theory is both necessary and sufficient to produce a given outcome, it can at the very least shed light on the plausibility (or implausibility) of a causal relationship.
85
In accordance with these guidelines, I drew from two established theoretical traditions—the dyadic democratic peace argument and a relational contracting perspective—with the goal of assessing their ability to account for the outcomes in both cases. For each argument I posited relationships between the independent and dependent variables. The congruence method enabled me to determine whether either theory offered predictions that were consistent with relationships observed in each of the selected cases. Evidence of consistency and inconsistency, in turn, helped me to identify future directions for Intelligence Studies theory development.
Identifying and testing for the observable implications of these two theories offered distinct advantages as well. First, it helped to guide my collection of primary source data by illuminating observations that were relevant (as well as irrelevant) to each theory in a given case. Second, testing for observable implications helped me to gain leverage over my small-n research problem. Despite using only two case studies, I was able to ‘increase my n’ by identifying multiple observable implications for each theory in each case (Keohane, King and Verba 1994, 220-221).
Measuring the dependent variable
Since the two theories examined in this dissertation purport to explain distinct aspects of a single phenomenon (depth of intelligence cooperation), it will be helpful to define more specifically how I measured each of these aspects. The dyadic democratic peace argument explains the depth of intelligence cooperation in terms of the relative frequency of contact between services and the granularity of intelligence exchanged. The relational contracting argument, by contrast, explains depth through
86
the lens of the structural commitments binding the two intelligence services together.
Bindingness
Frequency of contact: Meeting Opportunity Realization Rate (MORR). I identified several methods for capturing the variability of intra-WCID intelligence cooperation as a function of their frequency of interaction. The first is to simply document the number of days per month which face-to-face intelligence exchanges occurred. A second method is to tally the number of days per quarter in which face- to-face intelligence meetings (DM) occurred, and then divide that figure by the total number of calendar days that constitute each quarter (DQ), assuming that each calendar day represents an opportunity to meet and share intelligence. The resulting quotient provides a ‘Meeting Opportunity Realization Rate’ (MORR), which I define as the percentage of total calendar days during a given quarter in which both sides of a WCID gather together in the same physical location for any of a variety of purposes:
A relatively high MORR (approaching the score of 1.00 on the scale) is consistent with Wetzling’s description of “institutionalized information sharing among befriended services.” By contrast, a relatively low MORR (approaching the score of 0.0) is consistent with Wetzling’s description of either “no contact among foreign intelligence services” or “occasional information sharing among foreign services” (Wetzling 2005, 11). When meeting frequency was not documented sufficiently or explicitly in my selected data sources, I employed a fallback strategy
87
of measuring—on a monthly basis, in binary (yes/no) fashion—whether face-to-face bilateral intelligence exchanges were known to have occurred.
Granularity: Average maximum granularity of intelligence exchanged
(AMGIE). The mere occurrence of intelligence exchange is by itself an insufficient indicator of the intensity of bilateral intelligence cooperation. It is possible that some meetings may involve the exchange of more (or less) sensitive information than others. For instance, I may discover that the British Military Mission in Moscow exchanged order of battle (OB) information with the Soviet General Staff in one meeting, but two weeks later they engaged in a technical discussion of highly sensitive methods used to collect and exploit Axis military communications. Such distinctions can be captured by incorporating Clough’s notion of granularity. In an effort to define this concept more precisely, I conceptualized it in terms of three variables:
• Specificity captures the relative precision of the information being supplied. General information, such as operational summaries and estimates, will be scored with a value of 1. More precise information, such as the specific locations of enemy forces, will be scored with a value of 2.
• Sensitivity is an indirect function of plausible cover. A particular intelligence good becomes more sensitive from the supplier’s perspective when the range of plausible (alternative) sources from which it conceivably could have been derived diminishes. Items of low sensitivity (read: multiple plausible sources) will be scored with a value of 1. Items of high sensitivity (read: few plausible sources) will be scored with a value of 2.
• Acquisition costs refer to the initial expense borne by the prospective supplier in acquiring an intelligence good. Items that are relatively inexpensive to acquire, such as second-hand impressions of enemy morale, will be deemed relatively inexpensive (earning a score of 1). By contrast, items that are relatively expensive to acquire, such as details of Nazi Germany’s domestic chemical warfare program, will be deemed
88
expensive (thus garnering a 2).
It is worth noting that I do not allow for intermediate values in my scoring of these three attributes of granularity. This was a deliberate choice on my part. In an effort to systematize the coding process and ensure that incidents of intelligence exchange were coded consistently, I opted to conceptualize each attribute as the answer to a yes/no question. Thus, a particular item of intelligence exchanged between partners was either: general or specific; not particularly sensitive or highly sensitive; and cheap or relatively expensive to acquire.
Recognizing the potential for multiple intelligence exchanges occurring on the same day, I chose to select the highest scoring (read: most granular) intelligence good for each calendar day in a given quarter (DQ) and calculate the average of these maximum daily granularity scores (max(Gi)) over the course of each calendar quarter.
This provides the basis of a quarterly measurement of Average Maximum Granularity of Intelligence Exchanged (AMGIE):
Degree of structural commitment: bindingness. As discussed previously in
Chapter One, a weakness of Wetzling’s intensity continuum lies in its incorporation of two distinct concepts: frequency of intelligence exchange and the extent to which the two parties are bound in their commitment to one another. Since these concepts represent separate aspects of the dependent variable, it is important to disentangle them and devise separate measurements for each. Frequency of intelligence exchange
89
can be captured through the MORR measurement. Bindingness has been studied at great length by Weber, who defines it as a function of the member-states’ commitment to curtailing their sovereign discretionary powers through such mechanisms as arbitration procedures, decision-making structures and the integration of national militaries. Consistent with this view, a tightly bound relationship is one in which members face high exit costs and have limited room for maneuver. A weakly bound relationship, on the other hand, is one in which members face few significant exit costs and maintain their freedom of action (Weber 1997).
Figure 6. Weber's Continuum of Cooperative Security Arrangements
Weber’s cooperative security continuum, illustrated above in Figure 6, suggests that institutional bindingness increases as a security relationship moves from positions on the left toward those on the right. Locating and measuring those positions on the continuum is a more difficult proposition, however. As Weber does not define or distinguish clearly between formal and informal alliances (which comprise two of the three points on her continuum), we are left with only one type of
90
relationship – a confederation –whose position on the continuum can be plotted with any confidence. This conceptual imprecision is problematic, particularly when applied to the study of intelligence, where a broad range of conceivable relationships74 exists to the left of the ‘confederation’ position. How does one measure the bindingness of these relationships? In subsequent writings on this subject, Weber and Smith suggest a useful starting point:
We can often detect this deontological change in terms of rule-making when commitments change from a weaker formulation (“ought” or “should” or “shall”) to an explicit imperative (“will” or “must”). A more binding commitment also often involves greater clarity and formality in terms of the rule-specification; in other words, a general “rule of thumb” or “unwritten rule” becomes a specific, written rule (or even a law) as one moves along the continuum. Accordingly, rights and responsibilities become far more explicit and third-party dispute resolution, monitoring, and sanctions become more common (Weber and Smith 2004, 11-12).
When it comes to evaluating the bindingness of security relationships, clarity and formality serve as potentially useful distinctions. Clarity is defined in this study as the semantic precision governing a mutual commitment – perhaps manifested as the difference between two parties declaring a mutual commitment to one another’s defense, on the one hand, and agreeing to specific details about how that commitment will be carried out, on the other. Likewise, formality is defined as the degree of reputational investment that both sides have made in the relationship. Formality enables us to distinguish between a verbal agreement between principals and a written agreement that has been endorsed by both governments. In general, it follows that written agreements with sunset provisions and specific implementation language are indicative of a relatively binding relationship, whereas verbal agreements with very
74 Consider, for instance, some of the examples described in the beginning of Chapter One (p. 4).
91
general (or no) implementation language are less binding. This is illustrated in Table
1 below.
Table 1. Bindingness as a Function of Clarity and Formality of Commitment
The distinction between verbal and written agreements is probably less significant in the intelligence sphere than in the realm of general security cooperation.
This owes both to the highly private nature of bilateral intelligence exchange (that hidden “world within a world” of intelligence diplomacy described earlier by
Aldrich) and the corresponding reduction in public reputational investment made by the parties. If either party defects from a secretive intelligence sharing arrangement, the public costs associated with that decision will generally be less (e.g., in terms of international prestige) than they would be following defection from a publicly proclaimed alliance. Nevertheless, the private exit costs (e.g., in terms of loss of capability, forfeiture of fixed assets, etc.) could still be quite significant, which helps to explain why some intelligence partners explicitly “structure their commitments to minimize the chances and costs of defection by other participants” (Walsh 2010, 19).
92
Measuring the independent variables
Polity score. The dyadic democratic peace argument requires that one ascertain whether a WCID is fully liberal democratic or, alternatively, consists of one fully liberal democratic country and one fully autocratic country. These values can be ascertained by obtaining the polity scores (derived from the Polity IV dataset described above) of the UK, US, and USSR from 1940 through 1945.
Level of external threat and transaction costs. The relational contracting argument requires that one measure the levels of external threat and transaction costs facing the members of the WCID. Weber operationalizes the level of external threat as a function of the military strength and geographic proximity of a potential aggressor (Weber 1997, 331). This is premised on the long-held assertion that “states that are nearby pose a greater threat than those that are far away” (Walt 1985, 10). As seemingly straightforward as this relationship appears, it has proven much more difficult to measure in practice.
This is seen first in the attribute of military strength, which is presumed to have some potential bearing on the perceived likelihood of an enemy attack, along with the estimated damage that would result from such an attack, on one or both of the WCID members’ homelands. Military strength is typically captured with proxy measurements such as relative annual military expenditures and the number of standing military personnel. These measurements tend to “conceal gross inefficiencies” and often say very little about the ability of states to harness and project their power globally (Wohlforth 1999, 13).
93
Military strength is also problematic as a proxy because it lies in the eye of the beholder(s), which introduces the problem of misperception on the part of the threatened. Even assuming that a potential aggressor’s military capabilities could be accurately quantified, the threatened state must still grapple with the problems of uncertainty and intent. Such concerns can lead decision-makers to “see imaginary dangers… that critical observers regard as miniscule” (Jervis 1976, 372-373). It is therefore unlikely that one country’s military strength will present itself as an objective and immutable fact to all other countries in the international system.75
Geographic proximity is likewise presumed to shape the level of external threat because, per Walt, “states that are nearby pose a greater threat than those that are far away” (Walt 1985, 18). This is a seemingly more straightforward proxy to measure, given that the physical distance separating two countries is generally less subject to interpretation. The Correlates of War (COW) project was among the first research initiatives to apply standard units of measurements to geographic proximity, focusing specifically on direct contiguity relationships over land and water between states in the international system (Gochman 1991).
COW’s traditional measurements represent a sensible approach for dealing with contiguous dyads, but they are arguably less useful for dealing with pairs of non- contiguous countries aligned against a common non-contiguous enemy. For the cases selected in this study, only one stable contiguous relationship existed during the period of observation: the less than 400 miles of water separating Germany and the
UK from 1866 through 1945. In accordance with COW’s criteria, none of the other
75 As an alternative proxy, one might instead choose to measure the level of an adversary’s perceived military capabilities—but this too presents a number of additional methodological obstacles such as getting ‘inside the head’ of the prospective partner states.
94
relevant pairings – USSR/Germany, UK/Japan and US/Japan – can be measured, since by definition they are not instances of contiguity (Correlates of War Project
2007).
As I shall argue in the ensuing chapters, the role of external threat in my selected cases is arguably moot. While one could debate the extent to which the
USSR viewed some of the other Axis powers as external threats, by 1940 it appears that all three countries (US, UK, and USSR) had come to appreciate the imminent and mortal threat posed by the Axis Powers.76 Indeed, by this time all three countries had explicitly identified Nazi Germany and Japan as a common target of their respective intelligence efforts.
To the extent that parties involved in ad hoc cooperation are uncertain about their future prospects of cooperation and fear defection, that relationship can be said to have high transaction costs. As noted previously, Weber identities transaction costs as the second of her two independent variables, and she conceptualizes them as a function of three attributes: asset specificity, environmental uncertainty and heterogeneity between states. The first of these, asset specificity, provides a useful mechanism for measuring transaction costs. As one country invests a greater proportion of its asset-specific goods in an ad hoc intelligence relationship with another country, the former begins to fear the possibility and associated costs of the latter’s defection. States invested in such a relationship are inclined, per relational contracting theory, to abandon the ad hoc market in favor of a more formalized relationship that locks in the benefits of intelligence cooperation.
76 For example, prior to the US declaration of war in December of 1941, America's recognition of the Nazi threat was apparent in actions short of war, most notably the Destroyers for Bases Agreement of September 1940 and Lend-Lease Act of March 1941.
95
IR scholars tend to view asset specificity through the lens of traditional economic factors of production. Just as certain types of economic activities involve investments in highly specialized land, labor and physical capital, so too do cooperative security and intelligence arrangements among nations. Weber points to the examples of “highly trained men, specialized equipment, or specific sites that states otherwise would not invest in or could not readily shift to” (Weber 1997, 333).
Lake emphasizes the role of specialized divisions of labor and technological investments resulting in joint production economies, which in many cases cannot be realized by either party outside of the relationship (Lake and Powell 1999, 9). Walsh highlights instruments of “intelligence collection and analysis” focused on highly specific problems and not easily transferable to other targets or intelligence problems
(Walsh 2007, 157-158).
Not all assets invested in a foreign intelligence liaison relationships map neatly to a single side of the transactional ledger, however. Bilateral intelligence cooperation frequently involves personnel from both sides interacting directly with one another, as well as sharing and discussing sensitive information. It is often this diverse set of interpersonal relationships that constitute the truly asset-specific goods, rather than the “broadly-defined factors of production (labor, capital, and land) that feature in standard general equilibrium models of trade” (Hainmueller and Hiscox
2007, 3).
Determining whether a particular investment is ‘asset-specific’ requires that two questions be answered. First, can the supplier of that asset easily find other buyers? Second, can the buyer acquire the same goods from an alternative supplier?
96
In terms of Anglo-Soviet and Anglo-American intelligence cooperation during the
Second World War, the answer to both questions appears to be a definitive ‘no.’ It is difficult to think of any Allied military intelligence asset observed during this period—whether Russian Order of Battle data, raw US intercept of Japanese communications, or British cryptologic expertise—that conceivably could have been obtained from alternative sources or exchanged with alternative partners. I will expand upon this argument in the next two chapters.
Weber argues further that as levels of environmental uncertainty increase, actors that make highly asset-specific investments face high transaction costs, and thus have a stronger incentive to lock in the benefits of cooperation through more binding arrangements. Without offering an explicit definition of uncertainty,77 she measures this variable as ‘high’ or ‘low’ depending upon the number of Great Powers in the international system and the presence or absence of clear, consistent signaling between states (Weber 1997, 331-332). I argue that the level of uncertainty is high and constant in both the Anglo-Soviet and Anglo-American cases, given that the ‘fog of war’ prevents intelligence organizations from knowing the consequences of their actions. Likewise, the secretive nature of intelligence exchange precludes actors from signaling their true preferences in public—a critical mechanism that allows one’s counterparts to infer what type of intelligence organization it is dealing with (Morrow
1999, 86). I therefore treat uncertainty as an environmental constant in both case studies.
77 I will use the following definition of uncertainty: “the extent to which actors are not fully informed about others’ behavior, the state of the world, and/or others’ preferences” (Koremenos, Lipson and Snidal 2001, 778).
97
Weber’s final determinant of transaction costs is heterogeneity—a function of linguistic, religious, cultural, and political difference between countries. According to Weber, “the greater the degree of heterogeneity between countries, the greater the likelihood of language problems, misunderstandings, and disagreements, and therefore the greater the need for translation, arbitration, and costly coordination, which increase transaction costs significantly” (Weber 1997, 333). In sum, she expects to see greater institutional ‘bindingness’ between heterogeneous societies because the increased costs of doing business will compel such states to ‘exit the market’ and enter into a more institutionalized arrangement with one another.
Weber does not specify how this variable is to be measured, but an intra-
WCID measurement can be derived from a simple comparison of the common language of government and the political system (democracy vs. autocracy) of each
WCID member. Language differences, as Weber herself has indicated, can create an observable demand for translators and language training. Likewise, the effects of political heterogeneity may manifest themselves in the private deliberations of WCID participants—who may express reservations about sharing intelligence with an authoritarian government or, alternatively, comment favorably on the idea of engaging in intelligence cooperation with a fellow democracy (Ibid.).
Summary
This chapter began with a discussion of two IR theoretical traditions that I have adapted and leveraged to explain different features of the depth of bilateral intelligence cooperation: the dyadic democratic peace argument and relational contracting perspective. Following an articulation of testable hypotheses for both
98
arguments, case selection criteria were devised to facilitate the theory development and testing objectives of this dissertation. I ultimately selected two cases with highly similar background conditions: World War II-era intelligence cooperation involving the Anglo-Soviet and Anglo-American WCIDs. Testing procedures and measurement procedures were then developed for both arguments. With these methodological foundations in place, the study now turns to an examination of the cases themselves.
99
Chapter 4: Anglo-Soviet Case Study
Setting the Scene
The official war diaries of the British Military Mission to Moscow (aka 30
Mission78) suggest that there was no shortage of “very anxious and dangerous moments” that threatened to jeopardize the Anglo-Soviet wartime intelligence alliance against Hitler’s Germany. While it is tempting to view these episodes as simple byproducts of the immediate pressures and challenges of the war, one should not underestimate the extent to which past experiences also shaped British views of the Soviet Union as a foreign intelligence liaison partner (and vice-versa). Indeed, the creation of a military and intelligence alliance with the U.S.S.R. cut directly against what one 30 Mission official described as “the general legacy of suspicion and mistrust inherited from [anti-Soviet British] policies of the [previous] 25 years.” The details of that legacy are important, as they in many ways prefigured the contentious dynamics of the Anglo-Soviet intelligence alliance.79
Inauspicious beginnings
Anglo-Soviet relations did not get off to an encouraging start. Britain, having backed the anti-Bolshevik ‘White’ forces during the Russian Civil War, became a target of Soviet-sponsored sedition, propaganda and espionage efforts as early as the
1920s. While Moscow probably viewed other Western capitalist democracies (e.g.,
78 The British Military Mission to Moscow was known as 30 Mission because it was one of many numbered British military liaison missions deployed throughout the world. 79 Col. Exham to DMI, WO. War Diary of No. 30 Military Mission (hereafter 30 Mission War Diary) for period 1st to 31st July, 1945, Appendix L. TNA WO 178/27. July 15, 1943, 1-2.
100
Germany) as being more susceptible to communist revolution in the near term, the relative significance of Britain in these machinations should not be discounted. ‘The
Island’80 epitomized the very capitalist political order that the Soviet Union sought to overturn: a pre-eminent imperialist power that was home to both potential allies (an excitable labor movement) and enemies (a sizable ‘White Russians’ émigré population,81 not to mention the British Conservative Party) of communist revolution
(Payne 2004, 29; Malia 1995, 159; West and Tsarev 1999, 45).
At the same time as it was sponsoring subversive activities against the UK, the
Soviet Union was also pursuing normalized trade and diplomatic relations with
London (Bennett 1999, 69). By 1921, three disastrous years of War Communism had resulted in “an extraordinary primitivization of Russian life” in which factory, mine and agricultural outputs had dropped to levels well below their historical norms
(Malia 1994, 129; 143; Payne 2004, 296). Lenin’s New Economic Policy (NEP) sought to regain this lost capacity through the implementation of a mixed economy, but the success of that project hinged on the availability of capital that the young
Soviet state could not muster domestically (Malia 1994, 163).
Against this backdrop, a normalization of relations with the UK began to make sense for the Soviet Union. The British Government was in a position to forgive (or at the very least, restructure) significant portions of Russia’s pre- revolutionary war debt. Normalized relations would also afford Moscow access to
Western technology, credits and grain—assets that potentially could help the USSR
80 ‘The Island’ was the cover term used by the Foreign Department of the Russian Secret Police Organization (OGPU) when referring to Britain (West and Tsarev 1999, 47). 81 One of the most well-known and influential of these White Russians was Yuri Sabin, the former (Tsarist) Russian ambassador to the Court of St. James (Ibid., 44).
101
grow out of its economic doldrums. London saw potential value in normalization as well. The British Foreign Secretary speculated that Moscow was in such dire economic straits that it was prepared “to pay almost any price” for British assistance—including the “cessation of Bolshevik hostility” within Britain’s traditional sphere of influence, to include India (Glenny 1970, 65-66).
Bilateral negotiations ultimately led to the signing of the Anglo-Soviet Trade
Agreement in 1921. While the agreement stopped short of conferring formal British recognition on the USSR (this would come three years later under the First Labour
Government of Ramsay MacDonald), it did yield positive returns for Moscow. Most notably, Russia went on to enjoy a positive balance of trade with the UK and was relieved of its obligation to repay Tsarist war debt (Ibid., 81-82).
Normalization offered fewer benefits for London. Contrary to Soviet assurances of an end to “hostile action or undertaking against Britain … or the British
Empire,” the Moscow-based Communist International (Comintern) continued to sponsor subversive activities on both fronts (Bennett 1999, 7). Perhaps the only notable difference in Soviet behavior was seen in its increased use of proxies, which provided Moscow with plausible deniability of a broad range of illicit activities carried out on its behalf. Evidence of this was seen as early as the mid-1920s, when the USSR funneled money through the Communist Party of Britain (CPGB) and the
National Minority Movement (NMM) to foment British labor disputes (Flory 1977,
710-711). The UK Intelligence establishment managed to uncover some of these activities, but in most cases the British Government did not act against them because of the sensitivity of the evidence and the risks of information leaks to the public.
102
Two dramatic and highly publicized episodes also played an important role in shaping British views of the USSR. The first of these, the 1924 publication of the so- called Zinoviev letter, ignited a political firestorm in Britain through its depiction of an ambitious Soviet campaign of subversion and a seeming ignorance of (if not indifference toward) this activity by the MacDonald Government. The letter, ostensibly written by the President of the Executive Committee of the Comintern, called on the CPGB to agitate within Britain for ratification of the aforementioned
Anglo-Soviet trade agreement. According to the letter, ratification would enable the
Soviet Union to “extend and develop the propaganda of ideas of Leninism in England and the Colonies.”82 A copy of this document was leaked to the Daily Mail, which published it four days before British parliamentary elections (Bennett 1999, 1-4).
Questions surrounding the provenance and authenticity of the Zinoviev letter would later undermine its credibility as prima facie evidence of Soviet subversion.83
While it is debatable whether the letter had any tangible effect on the already weakened Labour Government’s margin of defeat,84 it seemed to confirm in the minds of many British Government officials what the Home and Foreign Offices had long suspected: that the USSR was pulling the strings of the CPGB, coordinating a campaign of communist subversion and meddling within the UK, and violating the
82 From Mr. MacDonald to M. Rakovsky. “Enclosure in No. 1.” History of the Zinoviev Incident. TNA CAB 24/168. October 24, 1924, 2. 83 Today the Zinoviev letter is widely regarded as a forgery. While the true author remains unknown, historians have posited a general story line in which White Russian Intelligence Services produced the document, which in turn fell into the hands of British Intelligence operatives who already suspected the Soviet Union of engaging in this type of behavior and “were on the look-out for opportunities to further the Conservative cause in Britain” (Bennett 1999, 92; West and Tsarev, 1999, 33-43). 84 It is worth recalling that the Government had already lost a vote of confidence at the time of the letter’s publication (Bennett 1999, 93).
103
spirit of the normalized diplomatic and trade relationship which it had pledged to pursue in good faith (Ibid., 54-55).
It was precisely these sorts of activities that led high-ranking British officials to conclude that the Soviet Union had become a “purely negative and destructive” force in international politics (Little 1988, 293). During the Second World War, a number of high-ranking 30 Mission officials would harbor similar suspicions about the willingness of their Russian counterparts to cooperate in good faith on such basic matters as routine information exchange and repatriation of escaped British POWs on
Russian soil.
If the Zinoviev letter constituted the first significant blow to Anglo-Soviet relations, then the May 1927 raid of the All-Russian Cooperative Society (Arcos) was surely the coup de grâce. The raid, conducted by the London Metropolitan Police, was organized in response to an allegation that a missing British War Department document had been secreted to the Arcos facility for photocopying by Soviet agents.
While the raid was unsuccessful in recovering the alleged document, it yielded what may have been the first publicly reported evidence of Moscow’s use of its official diplomatic and trade facilities in Britain for subversive purposes. British authorities seized over “250,000 pieces of incriminating evidence [from the raid], including secret ciphers, propaganda films and addresses of [Communist Party] members throughout the world” (Flory 1977, 716). This evidence, combined with secret information already in possession of the British Government, was cited as rationale for severing diplomatic relations with Moscow and expelling the Soviet diplomatic mission in 1927. Thus deprived of an official cover for its intelligence operatives
104
stationed in Britain, Moscow delegated responsibility for foreign intelligence- gathering to a vast and growing network of ‘illegals’ (Russians residing in Britain under assumed names and identities) (West and Tsarev 1999, 44-45).
The Arcos Raid may have had an impact on the conduct of Soviet communications security (COMSEC) practices as well. Until 1927, the Kremlin had been under the false impression that Britain was unable to process and exploit its encrypted diplomatic communications with Soviet personnel in London. That changed in the aftermath of the Arcos Raid, when the British Government — convinced that it had a limited window in which to act — selectively declassified and published secret Foreign Office documents in order to buttress its claims that London- based Soviet diplomats were involved in Russian espionage operations (Flory 1977,
719-720).
As the Arcos accusations and supporting evidence were made public, Moscow deduced that some of the incriminating information must have come from intercepted and decrypted USSR communications. Moscow responded by switching to a One
Time Pad (OTP) cipher system which reportedly “defied [British] cryptanalysis for some time” (McKay 1997). The Soviet Union’s reputation for effective counterintelligence – coupled with a capacity for exploiting the vulnerabilities of open and democratic government — would become a constant source of concern for
30 Mission, which feared that any indiscretions by its personnel in Moscow would be discovered by the Soviet Secret Police and intelligence services.
Notwithstanding Britain’s decision to renew diplomatic recognition of the
Soviet Union in 1929 and a very brief period of limited military and intelligence
105
cooperation during the Spanish Civil War, bilateral relations remained tense and confrontational (Smith 1996, 7-8). During the 1930s British diplomats began to attribute the rise of communist party membership in Western and Southern Europe to
Soviet and Comintern meddling (Little 1988, 293-294). Perhaps not coincidentally, the Comintern became a prime target of British signals intelligence collection efforts during this same decade (McKay 1997).
* * * * * * * * * * *
These troubled beginnings help to explain how an atmosphere of mutual mistrust and unease had instantiated itself in both countries on the eve of Hitler’s invasion of Poland. Grim reports of Stalin’s purges and Russia’s historic fickleness as a British ally—exemplified by the Bolsheviks’ abandonment of the Allied cause in
World War I and the more recent Nazi-Soviet Pact—had contributed to a highly negative impression of the Soviet Union among the limited pool of ‘Russia hands’ who would be assigned to 30 Mission (Smith 1996, 7; Searle 2008, 77). During the war these anti-Soviet prejudices would occasionally manifest themselves in private
Mission correspondence.
The Soviet Union, for its part, was uncomfortable with the prospect of capitalist intelligence officers operating openly on Russian soil and interacting with
Soviet intelligence personnel. Stalin continued to harbor suspicions of a Western capitalist plot to sell out Russia to Hitler. He seems to have been particularly suspicious of Allied initiatives (e.g., Operation Velvet) that threatened to create opportunities for unmonitored Anglo-Soviet fraternization—particularly in the
106
sensitive sphere of intelligence.85 Accordingly, the Kremlin sought to reign in the travels, activities and size of 30 Mission during its tenure in Moscow (Choi 2003).
The road to 30 Mission
The possibility of a Anglo-Soviet wartime intelligence alliance emerged as early as June of 1940, when the Churchill Government put out feelers to Stalin regarding the possibility of wartime cooperation against Hitler. The prime minister’s tepidly worded proposal, submitted via the British Ambassador to Moscow, called for the creation of an Anglo-Soviet arrangement to counter Nazi expansionism in Europe.
In exchange, Britain was prepared to maintain the status quo antebellum in its relations with Moscow.
Stalin declined this initial offer of assistance. Nevertheless, the overture was significant because it signaled Churchill’s willingness to set aside decades of anti-
Communist posturing and make common cause with Moscow against Hitler’s
Germany. Churchill again reached out to Stalin in April of 1941, this time by sharing
‘sanitized’ intelligence of an imminent German invasion of the USSR. In mid-June the British foreign minister went a step further, informing the Soviet Ambassador that
London was prepared to create and staff an inter-service military mission in Moscow, should Germany initiate hostile action against the USSR. The purpose of this mission
85 Operation Velvet was a UK initiative devised in August 1942 to protect the Caucasian oil fields on Russia's southern flank through the deployment of Anglo-American aircraft squadrons. It was also intended to serve as a gesture of goodwill toward the Soviets, who were upset with the Anglos’ continued delay in opening a second front in Europe. The initiative was ultimately torpedoed by Stalin, who “worried that U.S. and British aerial domination would tarnish Soviet prestige on the eastern front” and feared that the establishment of an allied air base on Russian soil would create additional opportunities for Anglo-Soviet interaction that might result in a “political contamination” of Soviet personnel (From Rear Adm. Miles to Gen. Wilson. 30 Mission War Diary, 1st to 30th November, 1942, Appendix J. TNA WO 178/26. November 28, 1942; Choi 2003, 151-152).
107
would be to coordinate the delivery of military supplies to Soviet forces and establish intelligence links with Soviet service element counterparts (Smith 1996, 13-18).
Later that month, the Soviet Ambassador assented to the British proposal on the condition that there would be ‘strict reciprocity’ in every detail. The first members of 30 Mission arrived in Moscow on 27 June.86 In early July a parallel
Soviet Mission was stood up in London. Thus it came to be that, within a matter of weeks following the initiation of Operation Barbarossa, an Anglo-Soviet intelligence alliance and military assistance effort was underway.
London was nevertheless uncertain whether its support would turn the tide of war in Stalin’s favor. During the early summer months of 1941, Britain’s Joint
Intelligence Council (JIC) had made consistently pessimistic predictions of Russia’s prospects in a Soviet-Nazi war. One JIC assessment concluded that Hitler’s forces would defeat the Soviets in less than two months. Given this bleak outlook, 30
Mission’s objectives were relatively modest. Its principal near-term goals were to prolong the Russian war effort (and thus “keep the pot boiling” against Hitler) and collect as much useful intelligence as possible on Germany and the Soviet Union. In the event of a Soviet collapse, 30 Mission planned to retreat to unoccupied parts of the country, gather intelligence and coordinate Russian resistance efforts (Smith
1996, 13-15).
Without the benefit of direct access to Soviet archival materials, it is difficult to obtain a parallel view into Stalin’s calculus for establishing a military supply and intelligence relationship with Britain. Given the depleted nature of the Russian
86 30 Mission War Diary, 21st June to 27th July 1941. TNA WO 178/25. June 27, 1941, 3-6.
108
military arsenal and the relative dearth of Soviet technological expertise during this period, it is possible that he viewed British military assistance as an equalizer in the fight against Hitler’s forces. The prospect of benefiting from the esteemed and well- sourced British intelligence services was probably a powerful incentive as well.
The Soviet Union might also have perceived an opportunity to extract certain concessions from London. On several occasions during the summer of 1941,
Molotov and the SGS sought assurances that the British would act immediately to
“relieve pressure on the Russian front”–both by launching counter-force air strikes in
Western Germany and orchestrating the sabotage of Romanian oil fields.87 This demand for a second front would become a consistent Soviet refrain during the war.
So long as it went unfulfilled, Moscow feared that it would be forced to shoulder a disproportionate share of the fighting among the Allied partners.
Description of intelligence cooperation machinery
30 Mission served as Britain’s central clearinghouse in Russia for military supply and intelligence exchange with the Soviet General Staff. Each service section was responsible for coordinating shipments of British military equipment (e.g., tanks, munitions and ammunition) to Russia and overseeing their proper use by the appropriate Soviet forces. Each section was also responsible for providing its
Russian counterparts with various types of enemy and technical information in exchange for operational intelligence from the Eastern Front. In addition, the British service sections were responsible for gathering information on the composition, disposition and overall performance of the Soviet military. Figure 7 illustrates how
8730 Mission War Diary, June-July 1941. TNA WO 178/25. June 30, 1941, 7.
109
this division of effort was implemented within a typical 30 Mission military service branch section, circa 1944.
Figure 7. Organizational Division of Effort for 30 Mission
With the exception of a brief period in late 1941 when the Mission was relocated to Kuibyshev during the Battle of Moscow, 30 Mission was headquartered in the Russian capital. Table 2 lists the rotating cast of senior British officers who filled the post of 30 Mission chief, who was responsible for overseeing the three military service branch sections (Army, Royal Air Force and Navy) and an administrative support staff.88 During the latter half of 1941 and early 1942, an
Accredited ‘Y’89 (Signals Intelligence) Representative was also attached to the
88 “British Military Mission to U.S.S.R. (Army Section).” 30 Mission War Diary, March 1944, Appendix DD. TNA WO 178/27. March 1944. 89 In the 1943 BRUSA Agreement, “Y” was defined as intelligence concerning the “intercepting, decoding, interpreting, classifying and dissemination of enemy (and neutral) communications, and the use of D/F [direction-finding] and other specialized apparatus for establishing locations and identities of enemy transmitters.” Special Intelligence, by contrast, was a term reserved for either “ciphers…placed in a special category, owing to their importance and
110
Mission.90 Taken together, these four units performed the bulk of Britain’s wartime liaison activities in Russia.91 Of these, the Army section appears to have had the most expansive and operationally significant liaison role. With most of the Eastern front fighting occurring on land, a significant proportion of related intelligence fell under the Army’s purview.
British Military Mission - Moscow (30 Mission) Chiefs of Mission Dates of service Lt. Gen. Sir Frank Noel Mason-Macfarlane (B.A.) June 1940 – May 1942 Rear Adm. G.J.A. Miles (R.N.) May 1942 – March 1943 Lt. Gen. Giffard LeQuesne Martel (B.A.) April 1943 – Feb. 1944 Lt. Gen. M.B. Burrows (B.A.) March 1944 – Nov. 1944 Adm. E.R. Archer (R.N.) Nov. 1944 – May 1945 Lt. Gen. Sir J.A.H. Gammell (B.A.) May 1945 – Sept. 1945
Soviet Military Liaison Structure – Moscow
SGS officers assigned to liaise with 30 Mission Dates of service
Lt. Gen. Golkiov June - July 1941 Maj. Gen. Panfilov July 1941 – Sept. 1942 Maj. Gen. Dubinin Sept. 1942 - Nov. 1943 Maj./Lt. Gen. Slavin Nov. 1943 - Aug. 1945
Chiefs of Otdel V.S. Dates of service Lt. Col./Maj. Gen. Evstigneev Aug. 1941 - Sept. 1944 Maj. Gen. Kutuzov Oct. 1944 - Aug. 1945 Source: WO 178/25, WO 178/26, WO 178/27, WO 208/1787. Table 2. Anglo-Soviet Liaison Structure and Key Personnel, 1941-1945
Most British meeting requests with the SGS were channeled through the
Soviet liaison element known as the Otdel. Direct contacts with the SGS were limited and tightly regulated. Among the few exceptions in this regard were the Nazi
difficulty of solution” (Special Intelligence A) or intelligence “derived from the solution of lower grade ciphers” that were often used tactically (Sims 1997, 31; 33-34). 90 “Charter for Lieut. Colonel Crankshaw (Draft).” TNA HW 61/37. July 1942. 91 Naval liaison activities during this period were relatively decentralized. British Naval Liaison Offices (BNLOs) were established in the south and north of the country. The northern BNLO element (based out of Archangel) worked closely with the Red Navy to run and secure British supply convoys.
111
OB exchanges, which occurred on a fairly regular basis from the summer of 1941 through early 1945. In those rare instances in which 30 Mission personnel were granted permission to visit Soviet forces on the Eastern front, it was not unusual for the Otdel to send along a commissar or staff officer to “act as a break on local commanders” who might otherwise be inclined to speak candidly with their British guests.92
30 Mission leadership came to view the Otdel as the principal impediment to
Anglo-Soviet intelligence exchange. British impressions of the organization were effectively summed up in 1942 by the head of 30 Mission’s Army Section, who faulted the Otdel for “imposing intolerable delays on the transaction of business, preventing personal contacts between the [British] Mission and the SGS, and generally exhibiting all of the worst characteristics of a 'post-office' department.”93 In internal correspondence, 30 Mission officials regularly lambasted Otdel personnel for their perceived incompetence, inefficiency and ignorance of substantive matters.
These impressions worsened throughout the war, eventually prompting the Chief of
30 Mission to protest in writing to his Soviet counterpart that the “officers of the
Otdel do not fully understand the meaning of the word liaison."94
92 From 30 Mission to Troopers. MIL 6078. 30 Mission War Diary, 28th May to 27th June 1942, Appendix F.F. TNA WO 178/26. June 26, 1942, 1. This practice continued well after Stalin’s official disbandment of the commissar system in October of 1942. 93 Exham to DMI, WO. 30 Mission War Diary, 28th June to 27th July, 1942, Appendix V. TNA WO 178/26. July 17, 1942, 1-2. 94 From Rear Adm. Archer to Maj. Gen. Slavin. “Requests of a Technical Nature.” 30 Mission War Diary, 1st to 30th April, 1945, Appendix F. TNA WO 178/27. April 13, 1945, 1.
112
Hypothesis Testing
As discussed previously in Chapter Three, two rival hypotheses will be examined in this chapter:
• The dyadic democratic peace hypothesis (H1) posits a relationship between the depth of bilateral foreign intelligence cooperation and the regime composition of a WCID. To wit, fully democratic dyads (those consisting of two democratic governments) are expected to reach and maintain higher levels of intelligence cooperation during a war than mixed dyads (i.e., those consisting of a democracy and an autocracy).
• The relational contracting hypothesis (H2), by contrast, posits variance between the bindingness of bilateral foreign intelligence cooperation and a combination of two factors: level of external threat and transaction costs facing WCID members. The higher the level of threat and/or transaction costs incurred, the greater the degree of bindingness is expected to be.
In this section I explain my rationale for choosing the 30 Mission War Diaries as my primary data source. I then measure values over the posited dependent and independent variables with a view to testing the relationships posited by each theory.
I also consider observable implications for each theory and assess the extent to which case outcomes cohere with (or alternatively, depart from) expectations.
Rationale for data selection
I recognized early in my research that no single data set would be able to cover all wartime intelligence interactions between the UK and USSR. The sheer number of exchange channels—e.g., military, diplomatic and political—made that impossible. In the absence of a master dataset, I settled upon the 30 Mission War
Diaries housed at the UK National Archives in Kew, England (WO 178/25; WO
113
178/26; and WO 178/27).95 In addition to their relative consistency and level of detail over the duration of the alliance, the diaries document the activities of an organization that served as the epicenter of the Anglo-Soviet military intelligence relationship during the Second World War. From its inception in 1941 through its dissolution in late 1945, 30 Mission served as Britain’s lead intelligence unit in the
Soviet Union. I was unable to identify any other single unit or element whose records and correspondence would better reflect the nature of Anglo-Soviet military intelligence cooperation during this period.
In none of the above respects did the 30 Mission records disappoint. I was heartened to discover, for instance, that nearly every month of the alliance was represented in the dataset. Most of the monthly records contain detailed records of 30
Mission’s inter-service activities in Moscow, as well as blow-by-blow accounts of its dealings with Soviet personnel and headquarters elements in Britain. From these records I was able to generate a dataset of 854 instances of contact between 30
Mission personnel and their Soviet counterparts from June of 1941 through August of
1945. My working definition of ‘instances of contact’ was deliberately broad. I recorded every reported instance involving at least one member of 30 Mission interacting with at least one Soviet counterpart in any of a variety of venues or modes of communication, to include meetings and site visits, as well as written and oral correspondence.
95 I obtained access to the records during a September 2008 research trip to the UK National Archives, as well as through the services of a London-based professional researcher whom I hired to take digital photographs of specific records that he then sent to me.
114
Testing the dyadic democratic peace argument
Measuring the dependent variable
For the dyadic democratic peace argument, I operationalize the depth of intelligence cooperation as a function of two key factors: frequency of contact and the nature and granularity of intelligence exchanged. These provide a longitudinal basis for charting the intensity of bilateral intelligence cooperation between the UK and USSR during World War II.
Frequency of contact. Despite the establishment and staffing of joint service military missions in both countries through war’s end, there is no evidence to suggest that the Anglo-Soviet WCID ever reached the high and sustained levels of intelligence cooperation observed in the Anglo-American WCID. On the contrary, after an initial honeymoon period during the summer of 1941, Anglo-Soviet intelligence cooperation diminished at a fairly steady rate. Following the termination of military operations in Europe, the Anglo-Soviet military intelligence relationship appears to have dissolved altogether (Smith 1996, 247).
This reduction in the frequency of Anglo-Soviet intelligence exchange is illustrated in Figure 8 and Figure 9. Figure 8 displays the number of monthly face-to- face meetings involving participants from both the SGS and the British Military
Mission in Moscow (aka ’30 Mission’) from 1941 through 1945. Figure 9 examines the relationship from a slightly different perspective, measuring inter-service military meeting frequency as a function of quarterly Meeting Opportunity Realization Rate
(MORR).96 Despite the aforementioned honeymoon period in early 1941 and two
96 MORR was defined earlier as the percentage of total calendar days during a given quarter in which both sides of a WCID gather together in the same physical location for any of a variety of
115
notable activity spikes in 1942,97 the overarching trend in both figures is one of diminishing direct interaction – with the 3rd Quarter of 1943 (July to September) serving as the proverbial ‘point of no return’ for bilateral intelligence cooperation.
The only sustained increase in Anglo-Soviet meeting activity appears to have occurred during a three-month period in 1943 just prior to the onset of the 3rd Quarter.
It was during this period that the SGS held a series of unusually detailed discussions with 30 Mission personnel on Russian military tactics and the organization of the
Soviet fighting forces. The Chief of 30 Mission, Lt. Gen. Martel, was encouraged by this unexpected increase in exchange. In private correspondence with the British War
Department, he cited these developments as possible evidence of Russia’s willingness to “take [30 Mission] more into their confidence.”98
purposes. 97 January and May of 1942 were the only months during the war in which the observed change in number of meetings (relative to the previous month’s total) exceeded the standard deviation (4.96) of the average observed monthly change (.02) throughout the war. January saw an increase of 12 meetings relative to the previous month and May saw an increase of 9. 98 Martel. “Report No. 2 from the British Military Mission in Russia from Lt. Gen. G. L. Martel.” 30 Mission War Diary, 1st to 30th June 1943, Appendix B. TNA WO 178/27. June 15, 1943, 3. In hindsight, this observation would prove to be premature. From August 1943 through war’s end, the average number of face-to-face meetings between 30 Mission and the SGS declined to fewer than six per month. By comparison, the average number of monthly meetings during the preceding period had been almost two times that amount — approximately 11 per month from June 1941 through July 1943.
116
Figure 8. Anglo-Soviet Intelligence Exchanges (Moscow), 1941-1945
How is one to account for Moscow’s uncharacteristic levels of contact with 30
Mission during this period? It is worth recalling that in April of 1943, the SGS had uncovered Nazi plans to launch a pincer attack on the Kursk salient later that summer.
Around the same time, British sources came across similar information about German intentions at Kursk and relayed the corroborating intelligence to Moscow—thus providing a short-term incentive for greater short-term cooperation with 30 Mission
(Mulligan 1987, 241). There is some evidence to support this position. In mid-June, for instance, General Martel noted that the Chief of the Red Army had “unusual[ly]” extended his personal thanks to 30 Mission for sharing information about the German forces. Such direct expressions of gratitude from high-level Soviet military leaders were exceedingly rare. In this case the remark may have been designed to nurture good will and encourage the continued supply of information concerning German activities and intentions on the Eastern Front.99
99 Martel. June 15, 1943, 1.
117
Figure 9. Anglo-Soviet MORR, 1941-1945
Nature and granularity of intelligence cooperation. The steady decline in face-to-face meetings between 30 Mission and the SGS provides one view into the state of Anglo-Soviet intelligence cooperation during World War II. By themselves, however, measurements of frequency do not sufficiently capture the depth of bilateral intelligence cooperation. For example, an overall decline in face-to-face exchanges could potentially obscure a deepening of cooperation in a single area of critical importance to the war effort. These risks can be mitigated by developing and incorporating measurements for the granularity of intelligence exchanged between the two parties.
To do this I first sought to identify a set of discrete intelligence goods that were exchanged in at least one calendar year quarter between SGS and 30 Mission, as observed in my dataset of 854 interactions. I then devised a crude ordinal coding scheme to measure the granularity of each topic. Consistent with the methodological
118
guidance outlined in Chapter Three, I measured granularity as a function of three attributes: specificity (general vs. specific), sensitivity (innocuous vs. sensitive) and its estimated initial cost of acquisition to the supplier (cheap vs. expensive). Table 3 below defines each type of intelligence good and the granularity values that I assigned to each.100
Table 3. Intelligence Goods and Estimated Granularity Measurements
I chose to measure granularity of intelligence exchanged in face-to-face exchanges, which constitute approximately 51% of the 854 interactions in my dataset.
100 For the vast majority of intelligence topics and sub-topics (21 out of 23) listed in the table, coding values for all three attributes were applied consistently across cases. Inconsistencies nevertheless surfaced in the coding of cases of ‘Chemical warfare’ and ‘Technical/Enemy.’ Intelligence goods falling into either of these categories occasionally varied dramatically in terms of their specificity, sensitivity and estimated costs of acquisition. Rather than attempting to ‘paper over’ these discrepancies by assigning them uniform values, I made a conscious choice to code items in these specific categories on a case-by-case basis.
119
I did this for two reasons. First, it provides a common unit of comparison with my other proxy variable (frequency of contact). Second, I have greater confidence in the accuracy of the underlying measurements, since most of the face-to-face exchanges in this dataset are captured in a greater level of detail than other forms of interaction
(e.g., letters and summaries of phone calls).
So as not to dilute granularity scores by ‘averaging out’ multiple instances of exchange occurring on the same day, I based my measurements on the average maximum observed value of granularity exchanged (AMGIE) for each day in the dataset. All told, intelligence goods were exchanged in face-to-face meetings on 376 separate calendar days (approximately 21 per quarter).
As illustrated in Figure 10, the granularity of intelligence goods exchanged between the SGS and 30 Mission appears to have followed a fairly consistent, downward trajectory—reaching its peak in early 1942 and declining in successive quarters from mid-1943 through war’s end. That said, AMGIE appears to have been particularly volatile from mid-1942 through mid-1943. Perhaps the single greatest contributor to this volatility was the sensitive matter of ‘Y.’ During the late summer months of 1942, 30 Mission’s newly accredited officer for ‘Y’ matters held a series of detailed discussions with the SGS on cryptologic efforts and radio direction-finding
(DF) against enemy communications. Just as momentum appeared to be building in this area, Moscow inexplicably put the brakes on all further ‘Y’ cooperation. By the end of the year, the SGS had informed 30 Mission that Moscow wished to discontinue Anglo-Soviet ‘Y’ cooperation altogether.101 While this particular avenue
101 From Lt. Col. Crankshaw to Chairman, ‘Y’ Committee. MIL.7755. TNA HW 14/60. December 2,1942, 1.
120
of cooperation did not dry up entirely, it never again reached the peak levels of activity observed during the Third Quarter of 1942.102
When expanding the population to include all instances in which intelligence was exchanged (representing a total of 639 calendar days, or approximately 36 per quarter), AMGIE assumes a relatively smoother distribution. As illustrated in Figure
11, AMGIE gradually rose toward its peak level of 3.0 in the first Quarter of 1941, dropped significantly from mid-to late-1942, rose briefly during the Third Quarter of
1943, and then tapered off sharply and conclusively thereafter.
Figure 10. Anglo-Soviet AMGIE, 1941-1945 (Face-to-Face Meetings Only)
102 The history of the ‘Y’ channel is discussed in greater detail on pages 135-137 of this chapter.
121
Figure 11. Anglo-Soviet AMGIE, 1941-1945 (All Encounters)
Measuring the independent variable
Testing the dyadic democratic peace hypothesis requires that we next measure the value of the independent variable and assess whether the observed values align as predicted by the theory. This can be done by gathering measurements of the polity2 values (derived from the Polity IV dataset) for the UK and USSR from 1940 through
1945. As expected, a comparison of these two countries reveals a pronounced difference in polity type. The UK consistently registers scores of 10 from 1940 through 1945, whereas the USSR consistently registers scores of -9. This places the
Anglo-Soviet WCID in the northwestern quadrant of the property space diagram displayed in Figure 5—an ‘ideal type’ mixed WCID, if ever there was one (Marshall,
Jaggers et al 2011).
122
The clear separation of subtypes in this WCID — a fully institutionalized democracy allied with a near-fully institutionalized autocracy against a common enemy — and the erosion of observed bilateral intelligence cooperation are congruent with the dyadic democratic peace hypothesis, so far as it goes. Nevertheless, these values by themselves do not explain how or why the level of observed Anglo-Soviet intelligence cooperation declined during the war. To gain such insights, we will need to reexamine the causal logic of this argument – the aforementioned regime recognition dynamic – and test for its observable implications within the case.
Testing for observable implications
One of the underlying principles of the dyadic democratic peace proposition, as it has been adapted here, is that democratic members of a mixed WCID are expected to harbor suspicion and enmity toward their authoritarian counterparts.
They are not expected to adhere to the norm of contingent consent because they fear the possibility of being exploited. Under these circumstances uncertainty becomes unbounded and there is on longer a predictable range of expected behaviors.
Democratic intelligence services in such a relationship will be especially alert to indications that their authoritarian counterparts are cheating.103 To the extent that these suspicions are confirmed (or are perceived as having been confirmed), democratic services will behave much as neo-realist theory suggests they should. To wit, they will become increasingly concerned with the relative gains of their partners
103 Cheating, also known as defection, may occur in any number of forms discussed in Chapter Two, to include deliberately manipulating shared intelligence, sharing information with unauthorized third parties, withholding relevant intelligence or specific forms of intelligence cooperation, and exaggerating the accuracy or significance of source claims (Walsh 2010, 9).
123
and will resort to coercive tactics in their retaliation against perceived imbalances of intelligence exchange.
Observable implications of the dyadic democratic peace proposition Contingent (1) The Soviet Union will exploit opportunities to tilt the terms of intelligence exchange permanently in its favor. consent (2) Britain will retaliate against the Soviet Union to rectify any perceived predictions imbalances in intelligence exchange. (3) The Soviet Union will fail to meet Britain’s minimum expectations of expected behavior as an intelligence partner: Bounded a) British intelligence personnel will have direct access to their Soviet counterparts. uncertainty b) Informational requests on matters concerning the common enemy will be predictions answered in a prompt and direct fashion. c) The Soviet Union will afford British intelligence personnel sufficient latitude to carry out their official liaison duties in Russia. Table 4. Dyadic Democratic Peace Argument: Anglo-Soviet Case Predictions
If this application of the dyadic democratic peace proposition is to find empirical support, we should see evidence of its purported causal mechanism (the regime recognition dynamic) at work within the case. While it is difficult to identify causal mechanisms through direct observation,104 they can be detected and measured indirectly through their observable implications. Table 4 lists three of these, each addressing a specific causal mechanism of the dyadic argument.
Case prediction #1: The Soviet Union will exploit opportunities to tilt the terms of intelligence exchange permanently in its favor. From the earliest stages of its intelligence alliance with the Soviet Union, 30 Mission stressed the necessity of maintaining balanced terms of exchange. In a late June 1941 meeting with General
Zhukov, senior Mission officers requested “reciprocal Russian intelligence for British use” in exchange for the intelligence and military supplies that Britain was already
104 Causal mechanisms are notoriously difficult to measure. As one scholar has noted, the observed relationship between an independent and dependent variable may well be “measurable and testable, at least to some degree” but the “territory in between [them]…is often a morass” (Gerring 2010).
124
providing.105 Later that fall, a visiting representative from Britain’s Government
Code and Cypher School (GC&CS) informed his Soviet hosts that “none of the
British material [that they had just discussed] would be handed to them other than on a basis of full reciprocity.”106 Well into the final year of fighting, 30 Mission would continue to push for reciprocity in its exchange of certain German and Allied technical information with the SGS.107
Unfortunately for 30 Mission, the SGS was seldom authorized to make any binding decisions or commitments on such matters. As noted previously, Stalin was concerned that the presence of UK personnel on Russian soil could trigger a ‘political contamination’ of the Soviet security services (Ulam 1974, 319 as cited in Choi 2003,
152).108 In an effort to mitigate these perceived risks to Soviet political stability, most decisions concerning intelligence matters were referred to the Kremlin, whose
“constant shadow” loomed over all variety of Anglo-Soviet exchanges.109 The extent
105 30 Mission War Diary, 21st June to 27th July 1941. TNA WO 178/25. June 28, 1941, 5. 106 From Gp/Capt. Blandy to Head of R.A.F. Section, GCHQ. “‘Y’ Liaison Visit to Russia by S/Ldr. Scott Farnie.” TNA HW 14/20. September 22, 1941, 1. 107 From 30 Mission to Troopers. MIL 2579. 30 Mission War Diary, 1st to 31st March 1945, Appendix I. TNA WO 178/27, 15 March 1945, 1; From Exham to Maj.-Gen. Kutuzov. “Requests of a Technical Nature.” 30 Mission War Diary, 1st to 31st March 1945, Appendix J. TNA WO 178/27, 15 March 1945, 1. 108 In an April 1944 letter, General Burrows reasoned that the Russians “dislike the idea of having a Mission on their doorstep” because it “means extra work for all their Secret Police organizations who are automatically bound to keep perpetual supervision of the movements of all foreigners; it means that many Russians who come into contact with the British officers and ORs must realize our higher standards of living at all levels; it means an inevitable increase in Black Market activities…” (From Lt. Gen. Burrows to DMI, WO. “30 Military Mission – Situation in Moscow on 15 Apr 44.” 30 Mission War Diary, 1st to 30th April 1944, Appendix O. TNA WO 178/27. April 15, 1944, 3). 109 30 Mission War Diary, 28th July to 26th August 1941. TNA WO 178/25. July 30, 1941, 2-4. The War Diaries contain numerous anecdotes highlighting 30 Mission’s difficulties in establishing social relationships with its Soviet counterparts. While the hand of the Kremlin is not always clearly visible in these episodes, the general progression of events – i.e., initial acceptance of an invitation, followed by an awkward eleventh-hour rejection – suggests a consistent pattern of intervention by a higher authority. A typical example is seen in the following account from May of 1943:
125
of the Kremlin’s influence is seen in two episodes from September of 1941. In a private meeting between General Macfarlane and General Panfilov, the latter noted that neither he nor the Chief of the Soviet General Staff, Marshal Shaposhnikov,
“were in a position” to provide their own impressions of fighting on the Russian front. According to Panfilov, all such questions were “for the [Soviet] Government” to address.110 Later that month, Captain Scott Farnie participated in a series of exchanges with the Soviet ‘Y’ Service in which “no Russian Officer [was able] to answer a question when it [was] put to him” and “everything [had to] be referred to the Kremlin for a decision.”111
The steady flow of military supplies and intelligence from the British, coupled with the stifling influence of the Kremlin and relative dearth of information provided by the SGS, raised concerns among a succession of 30 Mission chiefs that the British were falling into a position of relative weakness vis-à-vis the Soviets. These concerns surfaced relatively early in the tenure of the first 30 Mission Chief, General
Macfarlane. In August of 1941 he expressed frustration with the Russians’ tendency to “pick our brains technically…and get out of us as much material as possible” while
After General Martel’s lecture on 31st May …, he invited Colonel Muromtsev and several officers of the Otdel to dinner on 1st June. Colonel Muromtsev accepted; the other officers said they would find out whether their duties would permit them to come. Attempts to obtain a definite answer on 1st June failed until some time in the afternoon, when the Otdel telephoned to enquire what the dinner was in honour of. It was explained that it was an unofficial dinner, and was not in honour of anything. About 6.0 p.m. all the Otdel officers, in turn, rang up to say that they regretted that their duties prevented them from accepting. When, after some difficulty, Colonel Muromtsev was contacted by telephone, he said that he also now found that he would be unable to attend (30 Mission War Diary, 1st to 30th June 1943. TNA WO 178/27. June 2, 1943, 1.). 110 30 Mission War Diary, 27th August to 27th September 1941. TNA WO 178/25. September 8, 1941, 14-15. 111 From Blandy to Head of R.A.F. Section, GCHQ. September 22, 1941, 1-2.
126
at the same refusing to “tell us anything operational beyond providing a general picture sufficient to ensure our confidence in their capacity for resistance.”112
Macfarlane’s successor, Admiral Miles, later questioned the wisdom of
“giv[ing] the Soviets something for nothing” and argued that any continuation of this behavior would make 30 Mission “appear ridiculous” to its hosts.113 General
Burrows echoed that same sentiment over two years later when he advised the DMI to cease providing the SGS with British and American OB dispositions below the level of armies. In Burrows’ estimation, the Russians had no interest on this subject and 30
Mission was only “degrading [itself] by giving [such information] gratuitously.”114
While the relatively one-sided nature of enemy intelligence information flows was a constant source of frustration for 30 Mission, the British Government was generally untroubled by it. London was more interested in enabling the Red Army to wear down and ultimately defeat the common enemy. Symmetric intelligence- sharing (as on objective unto itself) was thus deemed less important than providing the SGS with whatever intelligence resources and military equipment it required to prevail against Hitler’s ground forces. This became apparent in late July of 1941, when the British Chiefs of Staff instructed Macfarlane to prepare for sharing deciphered German air codes without any “corresponding concessions” from the
Soviets. The Chiefs, apparently discounting damage that this might do to 30
Mission’s negotiating credibility with the SGS, argued that even “[i]f [the] Russians
112 From MacFarlane to C.I.G.S. MIL 234. 30 Military Mission War Diary, 28th July to 26th August 1941, Appendix G2. TNA WO 178/25. August 1, 1941, 1-2. 113 From Head of Mission to DMI. MIL 4843. 30 Military Mission War Diary, 28th May to 27th June 1942, Appendix K. TNA WO 178/26. June 3, 1942, 1. 114 From Burrows to DMI. MIL 1419. 30 Military Mission War Diary, 28th May to 27th June 1942, Appendix K. TNA WO 178/27. June 15, 1944, 1.
127
refuse to reciprocate they will be at least in a better position to defeat [the] common enemy.”115
London would endorse this view well through the Battle of Kursk, a campaign in which Moscow is generally credited with having seized the initiative definitively from the Germans on the Eastern Front (Mulligan 1987). In late September of 1943, for instance, the Director of Military Operations (DMO) allowed that while “it [was] certainly very annoying not to get more intelligence from the Russians, … this was … really a minor point so long as they [kept] the war going so well.” The DMO added that the Chief of 30 Mission was “in danger of losing his sense of proportion” with his continued complaints concerning the imbalance of Anglo-Soviet intelligence exchange.116
The 30 Mission War Diaries suggest that the SGS did not hesitate to exploit this British dependency to its advantage. As early as August of 1941, General
Macfarlane expressed frustration with the Kremlin’s policy of “giv[ing] us practically no operational news in amplification of official communiqués” and specifically refusing “…to let me visit the SMOLENSK front or even to mark a map showing the limits of the German thrust.”117 Further evidence of Soviet stinginess was seen in
Moscow’s tendency to share enemy OB information with the British only after it had been exploited for domestic political purposes. One of the first documented examples of this occurred in January of 1943, when the Chief of 30 Mission issued a written protest to his Soviet counterpart regarding Moscow’s decision to withhold potentially
115 “Dossier on Russian Liaison.” TNA HW 50/11. July 30, 1941, 1. 116 From DMI to DMO. TNA WO 208/1838. 28 September 1943, 1. 117 For C.I.G.S. from MacFarlane. August 1, 1941, 1-2.
128
useful OB information about the Stalingrad campaign. The SGS had sat on this information for over three months, apparently waiting for the occasion of the New
Year to include it in a ‘big splash’ Soviet press release. 30 Mission interpreted this as a cynical Soviet propaganda ploy to buttress Russian support for the war. Consistent with Allied precedent, 30 Mission had expected this information to be conveyed as soon as it became available through dedicated military-to-military intelligence channels.118
Moscow was no less stingy in its authorization of operational site visits to
Russian units on the Eastern Front. 30 Mission viewed site visits as an invaluable mechanism for obtaining firsthand information on enemy battle tactics and garnering
‘lessons learned’ from the Soviet fighting forces.119 The SGS, for its part, seems to have viewed them as both a nuisance and a potential security risk for impressionable
Soviet soldiers. Accordingly, the SGS took measures to limit the number of site visits granted to 30 Mission personnel.
These actions did not go unnoticed in London and Moscow, where the British
Chiefs of Staff and 30 Mission (respectively) were monitoring the bilateral intelligence cooperation ‘balance sheet’ very closely. An inter-service report concerning Britain’s military relations with the Soviet Union highlighted the
“manifest disproportion between the visits granted to 30 Military Mission and those given to the officers of the Soviet Military Mission and Soviet Trade Delegation
Engineers” in London.120 General Martel aired his frustrations directly to Soviet
118 30 Mission War Diary, 1st to 31st January 1943. TNA WO 178/27. January 1, 1943, 1. 119 Martel. June 15, 1943, 1. 120 From Brigadier Firebrace to Secretary, Chiefs of Staff Committee. “Inter-Service Report
129
Marshal Vasilievski in a December 1943 letter. According to Martel, Britain had granted 105 force unit visits to the Soviet Military Mission in London during 1943, whereas the SGS had granted only 5 to Army and RAF officers in Moscow during the same time period. In his letter Martel noted that “no British officer [had] yet been allowed to see any active operations on the Russian front.”121 Despite repeated efforts to coax and coerce122 the SGS into granting more site visits, 30 Mission personnel remained confined to Moscow. The perception of a ‘visit gap’ would persist through the final days of the war.123
Consistent with the first case prediction, the Soviet General Staff exploited the
British Government’s dependence upon the Soviet resistance as well as its tendency to tolerate significant imbalances in intelligence exchange. When 30 Mission requested greater reciprocity in intelligence matters, the SGS either ignored the requests or referred them to the Kremlin, where they often died. There appears to have been much less interest on the Soviet side in ‘keeping the game going’ to one another’s mutual benefit than in maximizing gains and minimizing expenses (in terms of sharing intelligence with the British). This put 30 Mission in the difficult position of developing an equitable bilateral exchange relationship with a partner that had no
on Our Military Relations with the Soviet.” TNA WO 208/1838. November 22, 1943, 3. 121 From Martel to Marshal Vasilievski. 30 Military Mission War Diary, 1st to 31st December 1945, Appendix N. TNA WO 178/27. December 31, 1943, 1-2. Further evidence of a ‘visit gap' was seen in an end-of-tour report filed by Lt. Col. Exham, who noted that “in spite of our repeated requests, no visits were permitted to Soviet tank units [over the proceeding 12 months] nor were we able to discuss their organization or tactical doctrine” (From Exham to DMI, WO. July 15, 1943.). 122 At one point 30 Mission went so far as to threaten the suspension of a planned visit to Italy (controlled at that time by Britain) for Admiral Kharlomov, hoping that this would compel General Slavin to authorize a visit to the Eastern Front. This tactic did not work (30 Mission War Diary, 1st to 30th November 1943. TNA WO 178/27, November 13, 1943, 3.). 123 From 30 Mission to Troopers. “Personal for DMI from Admiral Archer.” MIL 2931. 30 Mission War Diary, 1st to 31st May 1945, Appendix B. TNA WO 178/27. May 2, 1945, 1-2.
130
inclination – and was under no obligation or pressure – to reciprocate. This is in direct contravention of the spirit of contingent consent, wherein each party tolerates temporary gains by the other on the grounds that such gains will not be used to maintain permanent leverage.
Case prediction #2: Britain will retaliate against the Soviet Union to rectify any perceived imbalances in intelligence exchange. As noted previously, the frequency and granularity of Anglo-Soviet intelligence cooperation decreased dramatically following the turn-around at Kursk. Concurrently, 30 Mission began to detect a definitive change for the worse in its treatment by the Soviet Government. In a late September 1943 message addressed to the War Office, the Mission characterized its domestic and military relations with the Soviets as “bad” and “poor,” respectively. In response, 30 Mission called for the “application of every form of reprisal in domestic relations” as well as “reprisals in connection with military relations…applied equally by all three services” in Moscow.124 The British Chief of the Imperial General Staff (CIGS) promptly poured cold water on this idea, reminding the Chief of 30 Mission that his “task of improving relations in general
[was] of more importance than the acquisition of improved information, especially in a time when the Russians [were] fighting so hard and successfully.”125
In spite of this admonition, the Chief of 30 Mission was not prepared to let the matter drop. The following month, he elaborated upon his position in messages addressed to the CIGS and Deputy CIGS. General Martel allowed that the Mission’s
124 From 30 Military Mission, Moscow to the War Office. “For Chiefs of Staff.” MIL 11. TNA WO 208/1838. September 26, 1943. 125 From CIGS to Martel. 73930. TNA WO 208/1838. September 29, 1943.
131
“persuasive policy” had “met with some success till middle of July” but thereafter it had “gradually failed” despite having made “every effort…to recover lost ground.”
Citing the general lack of Soviet cooperation over the previous three months and the need to invoke a “strong line…[to] improve our general relations,” Martel urged the
CIGS to “stand up to [the] Russians now.” He called specifically for threatening significant cuts in the shipment of spare tank parts and limiting the number of meetings granted to Soviet liaison officers assigned to the Russian Mission and Trade
Delegation in London.126
Following consultations with the British Foreign Minister and the UK
Ambassador to Moscow, CIGS ultimately decided against instituting these reprisals on the grounds that the upcoming Moscow Conference might yet produce unspecified
“improvement[s] in [Anglo-Soviet] military relations.”127 This may have reflected
London’s calculation that the risks of Soviet defection in the intelligence sphere were acceptable, so long as the Red Army continued its successful prosecution of the war on the eastern front. Rejecting that line of reasoning, Martel and his successors continued to advocate taking a harder line “even at risk of temporary adverse effect on [the] war.”128 These efforts were similarly unsuccessful.
Even in the face of these rejections, 30 Mission continued to advocate for retaliation through private channels. 30 Mission drew a direct line between London’s
126 From 30 Mission to Troopers. “D.C.I.G.S. from Martel.” MIL 101; and “C.I.G.S. from Martel.” MIL 102. 30 Mission War Diary, 1st to 31st October 1943, Appendices D and D2. TNA WO 178/27. October 9, 1943. 127 30 Mission War Diary, 1st to 30th November 1943. TNA WO 178/27. November 5, 1943, 2. 128 From 30 Mission to Troopers. “Personal to V.C.I.G.S. from Martel.” MIL 712. 30 Mission War Diary, 1st to 31st January 1944, Appendix H. TNA WO 178/27. January 12, 1944, 1.
132
perceived policy of “continu[ing] to toady”129 and undesirable outcomes such as the rise of anti-British sentiment in Russia and the Soviets’ lack of cooperation in the repatriation of British POWs.130 This is seen further in General Martel’s consistent calls from mid-1943 through early 1944 for more negative threats (i.e., threatening to withhold certain forms of cooperation from the Soviets) in lieu of positive inducements (i.e., offers of quid pro quo). Concurrently, 30 Mission used increasingly direct and confrontational language in its dealings with the SGS. The
British Government remained unconvinced, arguing that the Anglo-Soviet relationship should be preserved at all costs so long as larger war objectives were at stake.
Here it seems that the causal logic of the dyadic democratic peace proposition did not unfold precisely as expected. Despite arguing forcefully and consistently for retaliation against the SGS, 30 Mission was unable to persuade London to abandon its policy of restraint and conciliation vis-à-vis the Soviets. In this case, the policies and preferences of the British Government trumped the parochial considerations of 30
Mission. Future studies may wish to consider additional scenarios under which the preferences of national governments and their subordinate intelligence services conflict—as well as the implications of these clashes for bilateral intelligence cooperation.
129 From Burrows to DMI, WO. April 15, 1944, 3-4. 130 During the last year of the war, 30 Mission noted on multiple occasions how difficult it had become to secure the cooperation of the Soviet General Staff in making arrangements for the repatriation of British POWs held in camps that were soon to be liberated by the Soviets (From Head of Mission to H. E. The Ambassador. “Proposed Combined Military Committee in Moscow.” 30 Mission War Diary, 1st to 30th September 1944, Appendix H. TNA WO 178/27. September 21, 1944, 1- 2; From 30 Mission to Troopers. “Personal Brigadier Firebrace from Brinckman.” MIL 2647. 30 Mission War Diary, 1st to 31st March 1945, Appendix L. TNA WO 178/27. March 21, 1945, 1).
133
Case prediction #3: The Soviet Union will fail to meet Britain’s minimum expectations of expected behavior as an intelligence partner. As seasoned Russia specialists and intelligence professionals, 30 Mission personnel had deep knowledge of the Soviet authoritarian political system and its potential implications for bilateral intelligence exchange. Upon their arrival in Moscow, they harbored no illusions about the largesse of their new intelligence partner. They focused instead on obtaining the ‘bare minimum’ necessary to carry out their assigned mission: reasonable access to their military service section counterparts, prompt responses to informational requests concerning the common enemy, and sufficient latitude to carry out their official liaison duties from within the Soviet Union.
(a) British intelligence personnel will have direct access to their Soviet counterparts. 30 Mission staked its early prospects for successful intelligence liaison on gaining direct access to Russian technical and subject matter experts in Moscow.
Within days of its arrival, each section head was tasked with “establishing contact with the [corresponding] Russian Service Ministr[y]” and assembling “all available intelligence ready to hand over as soon as contact had been established.”131 General
Macfarlane also made the case for direct access in a meeting with Soviet Foreign
Minister Molotov, where he emphasized the “immediate advantages to be gained by closer contact and cooperation between the Mission and the Russian Staff.”132
30 Mission personnel viewed these initial meetings with the SGS as an opportunity to bypass perfunctory elements of the Soviet liaison infrastructure (e.g., commissars and the Otdel) and directly discuss operational matters such as the status
131 30 Mission War Diary, 21st June to 27th July 1941. TNA WO 178/25. June 27, 1941, 4. 132 Ibid., 7.
134
of fighting on the Eastern front.133 This reflected a prevailing point of view within the UK military intelligence establishment viz. that liaison works best when representatives are free to engage in frank and free-ranging dialogue with their functional counterparts.134
Anecdotal evidence suggests that 30 Mission was seldom satisfied—and more often than not, frustrated —with the level and frequency of access that Moscow was willing to provide. From the Mission’s perspective, the Soviet liaison system and the
Kremlin represented the principal barriers to direct access. Both factors were on display during the late summer and fall months of 1942, when the newly accredited representative of the British ‘Y’ Services, Lt.-Col. Edward Crankshaw, returned to
Moscow.135 Crankshaw was tasked with “induc[ing] the Russians to co-operate with
[the UK] on ‘Y’ matters” and extracting more valuable Signals intelligence from the
Russians than they had supplied up to that point.136
During his first month back Crankshaw was relatively successful in gaining access to his Soviet counterparts. From late September through mid-October the SGS reversed itself and allowed him to share ‘Y’-related materials with the Red Army.
133 During one September 1941 meeting involving 30 Mission leadership and several high- ranking SGS officers, General MacFarlane asked his Soviet counterpart to provide his “full and frank” assessment of fighting on the Russian front and to disclose “the real state of affairs.” MacFarlane’s language in this encounter suggests an attempt to invoke the mutual bond of military professionalism to obtain additional information that presumably did not appear in materials that had been supplied by the Russians through formal channels up to this point (30 Mission War Diary, 27th August to 27th September 1941. TNA WO 178/25. September 8, 1941, 14.). 134 Such a view is consistent with General Martel’s observation that “it was difficult to put detailed questions [to his Soviet counterpart, General Dubinin] without having general discussions to lead up to them” (“Record of Meeting with General Dubinin: 23rd September, 1943.” 30 Mission War Diary, 1st to 30th September 1943, Appendix K. TNA WO 178/27. September 23, 1943, 1.). 135 Crankshaw had served a previous stint with 30 Mission as the Army Section’s ‘Y’ specialist from September of 1941 through June of 1942. 136 “Charter for Lieut. Colonel Crankshaw (Draft).” July 1942.
135
The Russians soon reciprocated on a number of fronts, as evidenced by their decision to authorize meetings with a variety of technical specialists, provide details on the
Soviets’ cryptologic capabilities, host bi-weekly meetings to deal with ‘Y’-related matters, and speak more freely about the Soviets’ Wireless Telegraphy interception efforts.137
Just as Crankshaw began to consider how he might exploit this newfound readiness to cooperate, the Soviets abruptly changed course. In mid-October the
Otdel inexplicably suspended Crankshaw’s access to Major Tulbovitch, who up to that point had been the former’s primary ‘Y’ interlocutor in Moscow. Crankshaw deduced from this that the Russians had “relapse[d] into [their] old bad ways” and his suspicions were confirmed at an early December meeting between Admiral Miles
(Chief, 30 Mission) and General Dubinin (Chief, SGS). It was at this meeting that
Dubinin stated with no further explanation: “The Head of the [Soviet] Department dealing with Wireless Intelligence considers that cooperation on this subject should be discontinued.”138 At this point Crankshaw saw no further prospects for improved
Anglo-Soviet cooperation in the area of ‘Y’ cooperation. Following his return to the
UK in February, substantive Anglo-Soviet cooperation in this area appears to have atrophied.139
137 From Crankshaw to Chairman, ‘Y’ Committee. MIL 7076. TNA HW 14/53. September 26, 1942; From Crankshaw to Chairman, ‘Y’ Committee. MIL 7087. TNA HW 14/53. September 27, 1942. 138 From Crankshaw to Chairman, ‘Y’ Committee. MIL 7290. HW 14/55. October 16, 1942; From Crankshaw to Chairman, ‘Y’ Committee. MIL 7755. TNA HW 14/55. December 2, 1942. 139 The ‘Y’ portfolio was assigned to Cecil Barclay in May of 1943. Barclay, who worked from the British embassy in Moscow, does not appear to have ever secured the degree of access that Crankshaw managed to attain briefly in the fall of 1942. According to an internal unpublished GC&CS review of this period, “the whole thing petered out” following Crankshaw’s departure “although there is nothing to indicate what line he [Barclay] took. It seems probable he never had another interview
136
Crankshaw’s experience during this period was typical of other 30 Mission officers who struggled to secure consistent access to their functional Soviet counterparts. During one stretch in the fall of 1942, the overall level of access to
Soviet officers had diminished to such an extent that letter-writing had become 30
Mission’s “only means of communicating with the Russians,” who for their part seldom responded.140 True to form, 30 Mission pinned much of the blame for this state of the affairs on the arcane Soviet liaison system. In a report characterizing the state of intelligence cooperation between 30 Mission’s Army Section and the Red
Army from May of 1942 to April of 1943, Col. Exham faulted the Otdel for
“effectively holding the Mission at arms-length from the Soviet staff.”141
Even when the Army Section was able to establish contact with a particular
Soviet section (e.g., the Soviets’ German Order of Battle section and Artillery
Directorate), the ensuing meetings were usually attended and recorded by a member of the Otdel staff. According to Exham, this arrangement ensured that the Army
Section’s “relations with Russian officers [did] not become too close or too friendly” and also prevented Soviet counterpart sections from making decisions or offering an opinion on any matter “without previous reference to some vague ‘higher authority’”
(Ibid.).142
after one of about 19 Nov[ember]…” (“Dossier on Russian Liaison.” November 1943.). 140 From Crankshaw to Chairman, ‘Y’ Committee. MIL 7661. TNA HW 14/59. November 24, 1942. 141 From Exham to DMI, WO. July 15, 1943, 7. 142 30 Mission attempted on several occasions to overcome these constraints by inviting their Soviet counterparts to social engagements, such as dinners and lectures. In most cases this strategy failed, generally because the Soviet participants were denied authorization to attend.
137
The situation worsened as fighting on the Russian front entered its final stretch. In an April 1944 situation report, the newly installed Mission chief stated that
“[c]ontacts with Russians of any class or type are forbidden – the only exceptions being a number of specially chosen ‘hacks’ who are naturally primed for the purpose.” He went on to assert that “Russians who strike up acquaintanceship with
British or Americans are very apt to disappear and we have evidence of [the Soviets meting out] punishments of up to eight years in labor camps for [those exhibiting]
‘familiarity with foreigners.’”143
(b) Informational requests on matters concerning the common enemy will be answered in a prompt and direct fashion. From its earliest days in country, 30
Mission was sensitive to the Soviet Union’s robust counterintelligence capabilities and its capacity for detecting any hint of duplicity on the part of officially declared
British intelligence personnel.144 30 Mission personnel were particularly careful about discussing and soliciting information concerning matters other than the common enemy (e.g., the internal situation in Russia and the structure and composition of the Russian military), lest these gestures be interpreted by Moscow as grounds for immediate expulsion.145
By going out of its way to respect these boundaries, 30 Mission hoped to pave the way for a robust exchange of information and perspectives on the common enemy and his capabilities, plans and tactics. High-level assurances from the Kremlin
143 From Burrows to DMI, WO. April 15, 1944. 144 From Miles to Gen. Wilson. November 28, 1942. 145 “Report on Visit by Officers of the British Military Mission to the Russian Front (South- west Front): 11th to 19th May, 1943.” 30 Mission War Diary, 1st to 31st May 1943, Appendix G. TNA WO 178/27. May 1943; From Exham to Head of Mission. 30 Mission War Diary, 1st to 30th September 1942, Appendix N. TNA WO 178/26. September 21, 1942.
138
certainly helped to reinforce this initial optimism. During one July 1941 meeting with high-ranking 30 Mission officials, Stalin reportedly promised General
Macfarlane a visit to the front and told Admiral Miles that he could soon visit the
Russian fleet.146
30 Mission nevertheless began to detect a widening gulf between these Soviet assurances and reality as the war progressed. Contrary to expectations, the SGS did not provide 30 Mission with regularly updated lists of identifications of units and formations of German forces in country. Instead, much of this information was accumulated and held by the Soviet Government until it could be incorporated into
Russian press releases designed to depict the Soviet war effort in a more favorable domestic light. British frustrations with Soviet information hoarding continued through the end of the war.147 By the summer of 1945, 30 Mission leadership was bemoaning the Soviets’ refusal to participate in technical exchanges with 30 Mission, its failure to acknowledge verbal and written correspondence from 30 Mission, and its provision of “unsatisfactory answers” in response to questions about German armies in the East.148
(c) The Soviet Union will afford British intelligence personnel sufficient latitude to carry out their official liaison duties from within Russia. The initial contingent of UK military intelligence personnel who traveled to Moscow in the summer of 1941 certainly did not expect to have free license to operate within
146 30 Mission War Diary, 21st June to 27th July 1941. TNA WO 178/25. July 12, 1941, 21-22. 147 “Record of Meeting: 18th January, 1944.” 30 Mission War Diary, 1st to 31st January 1944, Appendix L. TNA WO 178/27. January 18, 1944; From Troopers to 30 Mission. 64742. 30 Mission War Diary, 1st to 29th February 1944, Appendix Q. TNA WO 178/27. February 3, 1944. 148 From 30 Mission to Troopers. “Personal for D.M.I. from Admiral Archer.” 30 Mission War Diary, 1st to 31st May 1945, Appendix B. TNA WO 178/27. May 2, 1945.
139
Stalin’s Russia. As noted previously, War Diaries from this period abound with references to the long ‘shadow of the Kremlin’ and its chilling effects on bilateral intelligence exchange, scheduling of meetings, Anglo-Soviet fraternization and travel of 30 Mission personnel within Russia.
Mindful of Stalin’s suspicions and the possibility that Moscow could sever the nascent intelligence partnership without warning, 30 Mission went to great lengths to avoid antagonizing its Soviet hosts. This was seen above all in 30 Mission’s deliberate efforts to minimize the appearance of any intelligence collection efforts targeting the Soviet Union. While 30 Mission gathered and relayed information concerning the disposition and morale of Soviet fighting to London, it did so with great discretion—as evidenced by its reluctance to solicit this information directly from Soviet personnel during routine exchanges, lest such efforts be interpreted as domestic intelligence gathering.
Based upon its historical appreciation for the “incredibly efficient Russian internal security authorities” and their robust counter-intelligence capabilities, 30
Mission also advised London to limit circulation of its reports on internal Russian matters and omit all references to Soviet OB information and criticisms of the Red
Army.149 The possibility of this information getting back to Moscow and its potentially devastating effects on Anglo-Soviet intelligence cooperation was clearly a significant concern to 30 Mission.
149 “Report on Visit by Officers of the British Military Mission to the Russian Front (South- west Front): 11th to 19th May, 1943.” 30 Mission War Diary, 1st to 31st May 1943, Appendix G. WO 178/27. May 1943. This sentiment was underscored in a letter from Admiral Miles to General Sir H. Maitland Wilson: “We have, from bitter experience, adopted the slogan: Snooping doesn’t pay in Russia!” (From Miles to Wilson. November 28, 1942.).
140
These constraints left 30 Mission with limited room for maneuver.
Nevertheless, to the extent that 30 Mission personnel demonstrated respect for Soviet law and worked within the Otdel liaison framework, they had reason to expect a certain degree of latitude from the Russians. At the very least, 30 Mission expected minimal interference from Moscow in the execution of its day-to-day liaison responsibilities, such as establishing secure communications links with London, running transport convoys between the UK and the USSR (carrying mail, personnel, weapons and equipment), and obtaining exit and entry visas for departing and returning UK military personnel.
During its first two years in the Soviet Union, 30 Mission experienced relatively few setbacks in carrying out its routine functions and responsibilities. This began to change in May of 1943, when the Russian Foreign Office (RFO) started withholding relief visas from departing 30 Mission personnel stationed in Northern
Russia, where relations with the Russian Navy had been relatively strong. In effect, this meant that long-serving British personnel (e.g., those nearing the end of their rotational assignments) no longer had authorization to leave the country. Around the same time, the RFO stopped granting re-entry visas for British military personnel, including six named individuals who were deemed essential to the Mission’s efforts.
The cumulative effect of these actions was a progressive reduction in the size of the
UK’s Army and Air Force presence within Russia.150
As the summer wore on, 30 Mission encountered further problems in carrying out its routine liaison duties. In July, the Moscow civil government demanded that
150 From 30 Mission to Troopers. “Chiefs of Staff from Martel.” MIL 9161. 30 Mission War Diary, 1st to 31st May 1943, Appendix M. TNA WO 178/27. May 24, 1943.
141
two UK military personnel be tried in Russian courts for their alleged criminal conduct in Archangel and Murmansk. This was in contravention of established precedent among wartime allies. Up to this point the Missions had been entrusted to handle such matters internally. Moscow also took the unprecedented steps in August of imposing “delay and obstruction … over the routine formality of [administering] passes for [British] personnel in North Russia traveling to Moscow” and demanding the removal of named British personnel from Russia for their alleged entry and search of “premises occupied by Soviet citizens.”151
Collectively, these incidents had a chilling effect on foreign intelligence liaison. Unable to depart for home leave to the UK (lest they run the risk of not being allowed to return to Moscow at a later date), denied authorization to travel inside
Russia and incapable of evading Soviet censorship of in- and out-bound mail, 30
Mission found itself increasingly isolated and hamstrung. As General Burrows observed in September of 1944, conditions had deteriorated to the extent that it had become “extremely difficult for me and my Mission to obtain any collaboration from the SGS in carrying out instructions which I receive from the British Chief of
Staff.”152
Over time, these dashed expectations had a cumulative impact on 30
Mission’s interactions with and perceptions of its Soviet counterparts, as well as upon overall levels of bilateral intelligence cooperation. This is seen first in 30 Mission’s increased efforts to pursue reprisals against the Soviet Government for its perceived
151 From S. B. N. O. Archangel to Adm. Fisher. 30 Mission War Diary, August 1943, Appendix M. TNA WO 178/27. August 24, 1943. 152 From Head of Mission to H. E. the Ambassador. 30 Mission War Diary, 1st to 30th September 1944, Appendix H. TNA WO 178/27, September 21, 1944.
142
failure (and in some cases, outright refusal) to honor its intelligence commitments. It is also seen in the extent to which certain channels of information exchange (e.g., the
‘Y’ and German artillery discussions led by Crankshaw and Firebrace, respectively) suddenly and inexplicably ‘dried up,’ apparently at the Kremlin’s behest. 30
Mission’s preoccupation with the bilateral intelligence exchange ‘balance sheet’ and its corresponding concern that the Soviets were benefiting disproportionately from the relationship became increasingly evident. British frustrations were also evident in the highly critical statements and language employed by 30 Mission leadership in their direct dealings with the SGS and the Otdel over the last 3 years of the war. It was not uncommon during this period for Chiefs of 30 Mission to rattle off litanies of complaints during regularly scheduled meetings, as well as in letters directly addressed to Soviet leadership.
Testing the relational contracting argument
Measuring the dependent variable
Bindingness. The relational contracting perspective seeks to explain the bindingness of an intelligence relationship. This concept was defined earlier as a function of the clarity and formality153 of the parties’ expressed commitments to one another. By this measure, the Anglo-Soviet relationship appears to have lacked the formal and conceptual underpinnings of the Anglo-American wartime alliance, which as we shall see in the next chapter was anchored in the language of several cryptologic agreements.
153 Clarity was defined in Chapter Three as the semantic precision governing a mutual commitment; formality was defined as the degree of reputational investment that both sides have made in the relationship.
143
This absence of clarity was not a result of indifference or lack of intent on 30
Mission’s part. As early as the summer of 1941, General Macfarlane had lobbied
London for the creation of a joint Anglo-Soviet command structure. He argued that
“the time had now come to initiate inter-Allied Joint Planning” lest Britain give
Russia “the impression that [it] only intend[ed] to help them by the exchange of intelligence and the provision of munitions.”154 Colonel Exham made a similar argument over one year later in his annual Army Section report, calling for the creation of a Joint Charter that would govern the activities of the London and
Moscow Missions while at the same time ensuring that “both Mission[s] [w]ould have direct and easy access” to the appropriate officials in their respective host countries.155
These aspirations never materialized. This owed in part to the British
Government’s reluctance to assent to these proposals, as well as the general operating tendencies of the Soviet General Staff. As noted previously, SGS personnel were seldom given the latitude to make formal (i.e., written) commitments to the British on behalf of the Soviet Government. Most of their agreements tended to be of a more informal (i.e., verbal) nature and addressed ad hoc arrangements for information sharing, as opposed to policies governing longer-term intelligence cooperation. Over time, 30 Mission grew frustrated with Moscow’s perceived tendency to make joint commitments in principle while stopping short of formalizing and delivering upon them.
154 30 Mission War Diary, 27th August to 27th September 1941. TNA WO 178/25. August 31, 1941, 4-5. 155 From Exham to DMI, WO. Untitled Army Section report. 30 Mission War Diary, 1st to 31st December 1942, Appendix Q. TNA WO 178/26. December 23, 1942, 2.
144
Table 5 provides several examples of intelligence commitments made by the
UK and USSR towards one another during the war. Each example is mapped to its corresponding position in the bindingness matrix introduced in Chapter Three. This is by no means an exhaustive list of every agreement reached by the Soviets and
British during World War II. Nevertheless, it is generally representative of the types of bilateral agreements discussed in the 30 Mission War Diaries. With a few notable exceptions, most of these arrangements were reached in private settings without the direct involvement of the Kremlin.
Of those few arrangements for which a written agreement was produced, intelligence cooperation was never mentioned specifically. The Twenty-Year Mutual
Assistance Agreement of 1942 contains a provision for “affording one another military and other assistance and support of all kinds in war” against the Axis powers—but it does indicate whether intelligence constitutes part of the “other assistance.” Likewise, the 1942 Agreement on Technical Information addresses the sharing of Allied warfare technology but makes no mention of sharing information about the common enemy’s warfare capabilities. These examples suggest that the
Anglo-Soviet WCID was not particularly binding in terms of its observed clarity and formality (The Avalon Project 2008; Beardsley 1977).
145
Table 5. Bindingness Matrix for the Anglo-Soviet WCID, 1941-1945
Measuring the independent variables
Examining the Anglo-Soviet WCID through the lens of relational contracting requires that we next measure the observed values of the posited independent variables that are expected to shape its bindingness.
Level of external threat. In Chapter Three I highlighted some of the potentially intractable methodological obstacles surrounding the measurement of the level of external threat. These issues can be overcome (or at the very least, neutralized) if it can be established that the level of threat facing both WCIDs was more or less equal and constant, thus allowing it to serve as a controlled variable. I believe this condition is satisfied for both cases. The Anglo-Soviet and Anglo-
American WCIDs arose and persisted during the same general time period and were
146
aligned against the same set of enemies: the Axis powers. Perhaps the only notable distinction concerned the primary targets of intelligence gathering and sharing.
Whereas the 30 Mission-SGS partnership in Moscow focused primarily on Germany, the US SIGINT Agencies and their British counterparts in Washington (as we shall see in the next chapter) were concerned primarily with Japan.156
The importance of this distinction begins to break down upon further examination, however, given that both relationships were focused on a country against which each of the WCID member states had already declared war. By
Weber’s definition, the very presence of such wartime alliances renders the external level of threat ‘extreme’ (as opposed to ‘low’ or ‘high’) for both cases (Weber 2000,
19). Given this relative symmetry of the level of external threat, we can treat the level of external threat as a controlled variable for both cases.
Transaction costs. A key variable in many IR relational contracting arguments is transaction costs, which were defined in Chapter Two as the price of conducting ‘arms-length’ intelligence exchange in the global intelligence marketplace. To the extent that parties involved in ad hoc cooperation are uncertain about their future prospects of cooperation and fear defection, that relationship can be said to have high transaction costs. In Chapter Three I noted that transaction costs
156 In terms of the larger Anglo-Soviet and Anglo-American intelligence efforts, both Germany and Japan were countries of concern. For example, a thorough examination of the 30 Mission War Diaries reveals that the topics of Japanese ciphers and Order of Battle were occasionally discussed with the Soviet General Staff throughout the war—in some cases, well before Moscow formally declared war against Japan in August of 1945. And while Germany was seldom mentioned in the monthly SLU notes (which focused almost exclusively on Japanese matters), this was only because Germany was the primary focus of Anglo-American intelligence efforts based out of the United Kingdom at Bletchley Park.
147
would be operationalized in this study as two proxies: asset specificity and heterogeneity.
Social capital was one of the most significant and highly specific assets invested in the Anglo-Soviet WCID, at least insofar as the British were concerned.
The closed nature of Russian society, limited opportunities for interaction with the
Soviet military and a significantly scaled-back wartime intelligence effort against the
Soviet Union157 left the British Government and its intelligence services with few dedicated windows into Russia. Given Britain’s continuous demand for insights into
Soviet military thinking, the Red Army’s impressions of fighting on the Eastern Front and general conditions within Russia, it is not surprising that London came to view 30
Mission and its military-to-military liaison channels as assets worth retaining.
The intrinsic value of these social assets hinged directly on the active cooperation of the Soviet Government. Some economists refer to this condition as asset co-specificity (Hall and Soskice 2001, 17, as cited in Hainmueller and Hiscox
2007, 10-11). Three military liaison channels from the Anglo-Soviet WCID appear to have been particularly co-specific:
157 The official history of British intelligence in World War II states that GC&CS ceased all work on Russian codes and ciphers on 22 June 1941, the date on which Operation Barbarossa was initiated (Hinsley, Thomas et al 1979-1985, as cited in Smith 1996, 17). This statement may be technically accurate, but evidence from the UK National Archives suggests that Britain never fully suspended its collection and exploitation of Soviet communications. Within the British intelligence establishment, discussions continued well into the late summer months on such topics as “reducing the Russian section to a nucleus,” maintaining a skeletal wireless telegraphy effort from the British listening post in Sarafand (Palestine), and delegating continued monitoring of Russian radio communications to elements of the Polish Underground Army stationed in Stanmore, near London (From C. in C. Mediterranean to Admiralty. 1724C. TNA HW 14/19. September 8, 1941; Peszke 2005, 51-52). The British Directors of Intelligence eventually downgraded Russian exploitation efforts to a scaled-back ‘watch’ of wireless telegraphy from listening posts in Flowerdown and Cheadle. At least one dedicated British analyst was tasked with monitoring Russian naval traffic and keeping the Admiralty apprised as to “any drastic changes of method.” Polish units at Stanmore and Sarafand, for their part, were entrusted with maintaining surveillance of Russian radio communications and performing some degree of cryptanalysis (From A.G. Denniston to Multiple Parties. Untitled Memo. TNA HW 14/19. September 30, 1941.).
148
• The ‘Generals’ Channel:’ 142 meetings, June 1941 – May 1945. Out of sheer operational necessity, one of the first liaison channels to emerge in the Anglo-Soviet WCID was a regularly scheduled weekly meeting between the top-ranking officers of 30 Mission and the Soviet General Staff. These meetings provided an opportunity for both sides to review significant operational developments on the Eastern Front, as well as discuss and submit new requests for information, equipment and materiel. Despite its relative longevity as a Anglo-Soviet military liaison channel, the number and frequency of meetings tailed off significantly following the Battle of Kursk in July of 1943. While highly granular intelligence was not generally discussed or shared at the Generals’ meetings, the channel did serve a valuable coordination function and proved to be the most resilient of the Anglo-Soviet WCID.
• ‘The German OB Exchange Channel: 68 meetings, July 1941 – January 1945.158 One of the primary founding objectives of 30 Mission was to enable the sharing of information concerning the common enemy. Accordingly, both parties demonstrated a mutual interest in acquiring and sharing intelligence concerning the dispositions, locations and identifications of German force units operating in Western and Eastern Europe. By 30 Mission’s own judgment, the semi-regular German OB meetings served as a “very useful and regular contact” channel in which information exchange was “fairly free.”159 These positive views of the German OB channel began to sour as the war moved into its final year. This is seen in a reduced frequency of meetings followed by increased British complaints of the Soviets becoming “singularly secretive about their own and German Order of Battle.”160 Nevertheless, the German OB Channel was also among the most resilient of the Anglo-Soviet WCID.161
158 There was also a separate Japanese OB exchange channel, but it receives comparatively less mention in the War Diaries. Only 8 meetings are recorded for the period spanning 4 August 1941 through 12 September 1942. Most of the discussion in these meetings focused on information concerning Japanese intentions, strategy and future offensive operations. 159 From Exham to DMI, WO. Untitled final report as head of Army Section, spanning period of May 18, 1942 through April 4, 1942. 30 Mission War Diary, 1st to 31st July 1943, Appendix L. TNA WO 178/27. July 15, 1943, 9. 160 From 30 Mission to Troopers. “Following for Gen Ismay from Gen Burrows.” 30 Mission War Diary, 1st to 30th June 1944, Appendix Q. TNA WO 178/27. June 23, 1944; From Exham to Major-General Slavin. Untitled Letter. 30 Mission War Diary, 1st to 31st December 1944, Appendix F. TNA WO 178/27. December 6, 1944. 161 This is seen specifically in the ability of the channel to withstand frequent personnel changes during its first three years of existence. I noted no fewer than 8 unique pairings of Soviet and UK military officers whom together comprised the German OB channel: • Captain Birse and Colonel Gusev (28 July 1941 – 19 November 1941, 3 meetings); • Captain Birse and Major Masenstein (5 February 1942 – 9 March 1942, 2 meetings); • Captain Birse and Lt. Col. Postnikov (4 May 1942 – 25 May 1942, 2 meetings);
149
• The Enemy Signals Intelligence (‘Y’) Channel: 16 meetings, December 1941 – July 1943. The initial purpose of Signals Intelligence cooperation, from Britain’s perspective, was to “establish a full liaison with the Russian 'Y' Service and continue the exchange of information and documents.”162 In recognition of the promise of early exchanges and the potential intelligence gains for Britain, Lt.-Col. Edward Crankshaw was named in the summer of 1942 as Britain’s accredited ‘Y’ representative to 30 Mission.163 Despite achieving a significant breakthrough in the joint exploitation of German police communications and at one point meeting twice per week with the accredited Soviet ‘Y’ expert, the channel effectively closed in December when Moscow declared a unilateral suspension of ‘Y’ cooperation. While discussions continued on an intermittent and infrequent basis through the end of the war, they never again reached the levels and heights of early- to mid-1942.
As co-specific assets, none of these liaison channels could conceivably have been withdrawn and reallocated to alternate wartime uses. From London’s vantage point, each channel existed only to the extent that the Soviet General Staff was willing to acknowledge and tolerate it. Likewise, of the few assets listed in Table 6.
Asset Inventory for the Anglo-Soviet WCID: British Investments with a potential alternate wartime use,164 none could have been removed and reallocated without
• Captain Birse and Col. Khlopov (5 July 1942 – 5 August 1942, 2 meetings); • Capt. Chapman and Lt. Col. Postnikov (14 August 1942 – 21 November 1942, 8 meetings); • Capt. Chapman and Col. Motinov/Lt. Col. Zakhvatov (23 November – 13 February 1943, 4 meetings); • Capt. Chapman and Lt. Col. Lapkin (4 March 1943 – 5 July 1943, 9 meetings); and • Capt. Chapman and Major Pavlov (12 July 1943 – 27 January 1945, 40 meetings). 162 30 Mission War Diary, 27th August to 27th September 1941. TNA WO 178/25. September 11, 1941; From Blandy to Head of R.A.F. Section, GCHQ. September 22, 1941, 1. 163 Charter for Lieut. Colonel Crankshaw (Draft).” July 1942. 164 In principle, the Russia specialists detailed to 30 Mission could have been recalled from Moscow and reassigned to handle Russian liaison duties in London. Alternatively, these same individuals could have been detailed to Britain’s skeletal wartime intelligence effort against the Soviet Union. British military equipment leased to the Soviet Union could also have been put to alternate uses. The hundreds of Churchill tanks deployed to the Eastern Front, for instance, could (in principle) have been disassembled, boxed up and shipped to some other Allied theater of war for use in ground operations.
150
significant difficulty. Over time, the perceived fragility of each channel increased
Britain’s costs of doing business with the Soviet Union.
May asset What did the British Relative Practical Comparative be Asset Asset gain from its ease of uses beyond value of reallocated category examples investment of this asset’s current alternate to other asset? withdrawal application? use(s) activities? Updated information The Order of concerning the Battle (OB) dispositions of German Exchange forces & Russian on the Channel Eastern front. OB details and insights Social ‘Y’ Channel into Soviet Signals No N/A No N/A capital intelligence capabilities Operational intelligence The and occasional high- Generals’ level Soviet views into Channel fighting on Eastern front British A better-informed (and information, potentially more intelligence effective) Soviet war Intellec- assessments effort against Germany tual No N/A No N/A and ‘Y’ capital Potentially greater intercept amounts of Soviet shared with information concerning SGS Germany Military A potentially more Physical hardware effective Soviet fighting Moderate to capital supplied by force and weakened Difficult 30 Mission enemy Ability to engage with UK Russia Soviets in several specialists Yes Yes Less distinct channels Russian area Labor Difficult expertise Understanding of the and Soviet political system language capabilities Table 6. Asset Inventory for the Anglo-Soviet WCID: British Investments
The additional complicating factor of heterogeneity also increased the day-to- day costs of doing business. In addition to the aforementioned complications posed by regime heterogeneity (i.e., intelligence liaison officers from a democratic regime operating overtly within a fully authoritarian country), there was also the matter of a pronounced language barrier separating 30 Mission staff and their Soviet hosts. In
May of 1942, the SGS was beginning to demonstrate “definite signs of co-operation
151
in [Chemical Warfare (C.W.)] matters,” but the absence of a Russian-speaking C.W. expert at 30 Mission had resulted in “only written exchange of information [taking] place.” This in turn had led to “serious difficulties regarding translation of technical terms which [threatened to] produce misunderstandings or even dangerous inaccuracies.” In an effort to address this problem, 30 Mission recommended that the
British War Department limit its questions about C.W. matters to “short (repeat
SHORT)” inquiries that “call for short and definite answers.”165 Consistent with the relational contracting perspective, such linguistic incompatibilities should have provided 30 Mission with powerful incentives to pursue a more formalized (and perhaps hierarchical) intelligence relationship with the Soviet Union.
Testing for observable implications
As noted in Chapter Three, one of the central contentions of the relational contracting perspective is that two states engaged in ad hoc intelligence cooperation with one another will feel compelled under certain conditions to develop a more binding relationship. Building upon Weber’s anarchic argument, Walsh’s willful hierarchy argument offers an explanation of how and under what conditions this bindingness may materialize. In cases where the expected benefits of intelligence cooperation are high, at least one party fears that the other will defect, and there exists a clear power disparity between the two parties, a hierarchical arrangement will be created in which the dominant party is empowered to “manage and oversee some of the other participant’s activities” (Walsh 2010, 17-20; 24-25). However, before we can test for the observable implications of this argument within this case, we must
165 From 30 Military Mission to Troopers. MIL 4733. 30 Mission War Diary, 28th April to 27th May 1942, Appendix K.K.2. TNA WO 178/26. May 24, 1942.
152
first determine whether the Anglo-Soviet WCID satisfies these preconditions. We can accomplish this by addressing three basic questions.
(1) Did both parties believe that intelligence cooperation against the common enemy would be fruitful? The British certainly did, as evidenced by their repeated efforts to cultivate better relationships with various subdivisions of the SGS. Soviet views on this matter are more difficult to assess, but the fact that Stalin agreed to the unprecedented establishment of a western capitalist inter-service military mission
(with overt intelligence functions) in Moscow suggests that he too saw value in this arrangement.
(2) Did at least one of the parties have reason to suspect the other of defecting? Given the troubled history of Anglo-Soviet relations prior to the onset of the war (in particular, Britain’s perception of Russia as an unreliable ally and Stalin’s reported fears of the ‘contaminating’ effect that British intelligence officers could have on their Soviet counterparts), there can be little doubt that each country harbored a certain degree of distrust toward the other.
(3) Was there a clear dominant and subordinate entity in this relationship?
While the Soviet Union had certain absolute material resource advantages over the
UK on the eve of war,166 these quickly eroded in the face of horrendous losses inflicted upon the Soviet economy and military force levels by the invading Axis forces (Hill 2007, 787). Likewise, Moscow lacked the domestic capacity for producing the sort of high-technology and high-quality equipment required to stem
166 Harrison finds that Soviet wartime GDP exceeded British GDP for all years except 1942 and 1943 (Harrison 1998, 10). The Soviet Union also had significant advantages on the eve of war in terms of the size of its civilian population and standing military forces.
153
the German offensive and regain the initiative by itself.167 Given its bleak outlook at this time and a lack of alternate suppliers, Moscow was arguably in no position to refuse Britain’s offer of assistance. If there were a dominant party in the Anglo-
Soviet WCID during this period, then it was almost certainly the UK.
Observable implications of the relational contracting perspective
(1) The dominant party will assume some measure of control over the subordinate party’s intelligence services.
(2) In exchange for its cooperation, the subordinate party will receive resources from the dominant party that exceed the value of shared intelligence.
(3) The dominant party will pursue any of the following measures in its dealings with the subordinate party to reduce the likelihood of defection: (a) limit intelligence sharing to specific topics of common interest; (b) clearly specify the types and/or topics of intelligence that will (and will not) be shared; (c) monitor the subordinate state’s compliance with sharing arrangements; (d) punish any reneging by the subordinate party; and (e) restructure and/or encourage reform of the subordinate state’s intelligence services, such that they are more inclined to cooperate.
Table 7. Relational Contracting Argument: Anglo-Soviet Case Predictions
Having determined that all three preconditions of the willful hierarchy argument have been satisfied, we can now test it further by considering some of the observable implications of this theory within the Anglo-Soviet WCID. These appear in Table 7 above.
Case prediction #1: The dominant party will assume some measure of control over the subordinate party’s intelligence services. At no point in the history
167 30 Mission was aware of this advantage and privately viewed its technological strength as a bargaining chip in its early dealings with the SGS. Colonel Exham hinted at this in a September 1942 memo, in which he assessed the anticipated value of British Signals intelligence to the Soviet General Staff: “I am sure [Crankshaw] has every chance to establish a valuable organization, because he has material which the Russians want badly. He is one of the few soldiers here with good weapons in his bag” (From Exham to Head of Mission. Untitled report. 30 Mission War Diary, 1st to 30th September 1942, Appendix N. TNA WO 178/26. September 20, 1942, 2).
154
of the Anglo-Soviet WCID is there any evidence of the UK attempting to assert direct control over Soviet intelligence activities. While Britain initially considered using 30
Mission to coordinate Russian resistance activities from behind enemy lines, this was strictly a secondary (fallback) strategy to be employed in the event of a sudden Soviet collapse. The primary objectives of 30 Mission were to prolong and augment the
Russian war effort against Hitler (Smith 1996, 13-15). To the extent that the Red
Army remained in the fight, London had no interest in granting 30 Mission any of the coercive levers typically wielded by a dominant party. Indeed, the Foreign Office,
CIGS and War Office were generally quite pleased with Russia’s performance on the
Eastern Front—particularly after the turn-around at Kursk.
Even in the face of sharp objections from successive 30 Mission chiefs,
London refused to endorse any actions or behaviors—to include the introduction of quid pro quo bargaining postures in intelligence matters—that would detract from the
Russians’ battlefield performance.168 This effectively neutralized any leverage that
30 Mission might have had as the UK’s principal conduit for intelligence exchange and military supplies with the Soviet Union. In fact, none of the hierarchical arrangements described by Walsh – to include direct management of the weaker party’s intelligence service, supervision of its collection activities, or vetting of its sources and intelligence personnel – ever materialized in the Anglo-Soviet WCID
(Walsh 2010, 20).
The willful hierarchy argument offers two possible explanations for why the expected outcome did not occur in this case. The first is that at least one of the
168 From Firebrace to DMI. Untitled note. TNA WO 208/1838. September 28, 1943; (Searle 2008, 70).
155
parties did not believe that hierarchy could be procured at an acceptable cost (Walsh
2007, 153). The Soviet Union may have held such a view, given Stalin’s longstanding suspicion of British motives and his related fears of Soviet forces being politically contaminated by Allied intelligence officers. The prospect of a western capitalist military element directing Soviet intelligence activities was most likely a non-starter for Stalin. Indeed, far from entering into a willful hierarchy with the
British, he proscribed direct interaction between 30 Mission personnel and Soviet intelligence officers in a variety of settings.
If the above explanation were true, however, we should expect to see evidence of similar concerns affecting other intelligence alliances between democracies and autocracies. Such evidence is hard to find. There are numerous examples of regimes with negative polity scores (e.g., South Vietnam, Jordan, Pakistan under Musharraf and Mubarak-era Egypt) entering into “hierarchical intelligence-sharing relationships that give the United States some ability to directly monitor and control its partners’ intelligence activities” (Walsh 2010, 113-114). The willful hierarchy view is silent on why so many authoritarian regimes would find these costs acceptable while others deem them cost-prohibitive.
The second possible explanation is that the absence of a dominant party precluded the creation of a willful hierarchical relationship in the first place. Walsh notes that hierarchy tends to occur only when one participant’s influence and contribution to the WCID is more important to the success of the joint venture than the other’s (Ibid., 18). This role typically falls to the dominant party, given its material resource advantages and ability to credibly threaten abandonment of the
156
relationship. If in fact there was no dominant party in the Anglo-Soviet WCID, then a hierarchical relationship should not have materialized.
There might be some merit to this explanation if we were to restrict our focus exclusively to the post-Kursk period, when Moscow seized the initiative against the
Nazis and became increasingly less reliant on the British. The argument falls apart, however, if we expand our consideration to cover the entire period of the alliance. As noted above, the USSR was at a clear resource disadvantage vis-à-vis- Hitler’s Army during the crucial period of the First Moscow Protocol, spanning roughly November
1941 through June 1942 (Munting 1984; Langer 1979). It is also worth reiterating that the UK assumed the role of “senior [Allied] partner in the provision of aid to the
Soviet Union” during this period, prior to any significant involvement by the United
States (Hill 2007, 782). As a key supplier of intelligence and logistical support to the
Soviet Union, the UK was for all practical purposes the dominant party in this relationship. The second explanation therefore fails to convince.
Case prediction #2: In exchange for its cooperation, the subordinate party will receive resources from the dominant party that exceed the value of shared intelligence. From the moment that the idea of a British inter-service military and intelligence mission was first broached with the Soviet Union in June of 1941, the
UK made clear its willingness to barter outside of traditional intelligence channels.
That message was reinforced within days of 30 Mission’s arrival in Moscow.
157
True or False Claim for the Anglo- Assessment Soviet WCID? Intelligence cooperation was limited to topics concerning the common enemy, with (a) Sharing is limited to specific emphasis on OB intelligence. 30 specific topics of common True Mission pushed for the inclusion of interest and around which additional topics (e.g., enemy ‘Y’ and both parties’ interests align. Soviet battlefield dispositions), but met with only sporadic and limited success. Despite infrequent verbal commitments to (b) The parties clearly share specific types of intelligence (e.g., at specify the types and/or the weekly Generals’ and OB Meetings), False topics of intelligence that the UK and USSR never managed to reach will be shared. a formal agreement governing intelligence cooperation in specific spheres. Such monitoring did not occur because an explicit Anglo-Soviet intelligence sharing (c) The dominant state arrangement was never reached. In fact, monitors the subordinate False there is evidence that 30 Mission actively state’s compliance with discouraged all forms of monitoring and sharing arrangements. intelligence gathering against the Soviets, given the latter’s reputation for detecting it. 30 Mission argued passionately for a policy of harder bargaining with the SGS. General Burrows went so far as to urge Churchill’s top military adviser to “hold up (d) The dominant state sufficient information from the Soviet introduces explicit False Mission in London to make them punishments for reneging. squeal.”169 As noted above, however, London declined to authorize retaliatory measures on the grounds that they might jeopardize Anglo-Soviet relations. 30 Mission leaders initially supported ideas (e) The dominant state such as creating a Joint Planning Staff and restructures and/or persuading the Soviets to replace the Otdel encourages reform of the with a “charter to cover the activities of our subordinate state’s False respective Missions in London and intelligence services, such Moscow” (WO 178/25, September 1941; that they are more inclined to WO 178/26, December 1942). London did cooperate. not find these arguments compelling. Table 8. Expected Anglo-Soviet Defection-Avoidance Strategies
169 From 30 Mission to Troopers. “Following for Gen Ismay from Gen Burrows.” 30 Mission War Diary, 1st to 30th June 1944, Appendix Q. TNA WO 178/27. June 23, 1944.
158
In response to several Soviet requests in early July for an intensification of air, naval and bomber attacks on German concentrations in both Russia and Western Europe, 30
Mission instructed its Naval and Air experts to “immediately work out the details.”170
Case prediction #3: The dominant party will attempt to reduce the likelihood of the subordinate party’s defection through any of a variety of measures. Having established that the Anglo-Soviet case meets the pre-requisites for
Walsh’s willful hierarchy argument, we may now consider some of its additional observable implications in this case. There is perhaps no better set of these than
Walsh’s list of “five ways [that dominant states can] minimize the chances and costs of defection by other participants” (Walsh 2007, 162). These claims are presented and tested in Table 8 above.
Conclusion
Despite a troubled history rooted in mutual distrust, the UK and USSR formed an alliance against Hitler’s Germany for the greater part of five years. Intelligence exchange played a significant role in this relationship. Nevertheless, with the exception of an initial honeymoon period and the several months leading up to the
Battle of Kursk, cooperation in the intelligence sphere never lived up to 30 Mission’s initial aspirations. This was evidenced by 30 Mission’s steadily diminishing degree of interaction with the SGS, as well as in the absence of a formally binding agreement to facilitate bilateral intelligence exchange.
The two exploratory hypotheses examined in this chapter addressed slightly different but related aspects of the dependent variable, depth of intelligence
170 30 Mission War Diary, 21st June to 27th July 1941. TNA WO 178/25. July 2, 1941, 9.
159
cooperation. The dyadic democratic peace argument made predictions about the anticipated degree and granularity of intelligence cooperation, as mediated through the regime recognition dynamic. Consistent with our expectations for a mixed
WCID, the SGS violated the norm of contingent consent by demonstrating no interest in ‘keeping the game’ going to one another’s mutual benefit. Likewise, by failing to adhere to any of 30 Mission’s bare minimum expectations of liaison partner behavior, the SGS created an environment of unbounded uncertainty. 30 Mission was unable to determine what constituted ‘acceptable behavior’ for its fickle foreign intelligence liaison partner.
Perhaps the only unexpected outcome in this case was 30 Mission’s failure to retaliate against the SGS. This had less to do with 30 Mission’s own preferences, however, than it did with the British Government exercising its prerogative. As much as 30 Mission saw redemptive value in (and argued passionately for) employing coercive measures in its dealings with the SGS to “make them squeal,” London felt otherwise. So long as the Red Army continued to hold its own against Hitler’s forces on the Eastern Front, the British Government was unwilling to authorize any actions or behavior by 30 Mission that might risk poisoning the relationship—regardless of its impact on bilateral intelligence cooperation.
The relational contracting argument made predictions about the bindingness of intelligence cooperation, as mediated through the level of external threat and transaction costs facing the participants. Here the relationships between the dependent and independent variables did not play out as expected. Despite fairly significant transaction costs and a high level of threat (treated here as a control
160
variable), 30 Mission and the SGS never reached any binding agreements to govern intelligence cooperation. Even so, a handful of relational contracting predictions did bear true. For instance, 30 Mission provided the SGS with a net package of goods that clearly exceeded the value of the intelligence that was being exchanged—perhaps as a means of reducing Moscow’s incentive to defect. Both parties also seemed to focus on a select subset of topics (e.g., German OB) throughout the war, thus suggesting a degree of specialization that is consistent with the relational contracting perspective.
These examples notwithstanding, 30 Mission at no point attempted or even came close to assuming hierarchical control over the SGS intelligence services, as suggested by the willful hierarchy view.
Of the two theories examined in this chapter, the dyadic democratic peace proposition seems to do a better job of explaining the Anglo-Soviet case outcomes.
The relationship between the dependent and independent variables is consistent with my predictions, and nearly all of the observable implications of this explanation were corroborated. Most of the claims derived from the relational contracting perspective, on the other hand, were not corroborated in the case data. The next chapter will explore how well the two theories perform against an intra-democratic WCID: the
Anglo-American relationship.
161
Chapter 5: Anglo-American Case Study
Setting the Scene
In May of 2011, US President Barack Obama and UK Prime Minister David
Cameron wrote that ties between their governments constituted “not just a special relationship, [but] … an essential relationship (Obama and Cameron 2011).” The official joint statement was very much in keeping with a 65-year tradition in which
British and American leaders have ritualistically invoked the ‘Special Relationship’— a term first popularized by Churchill in his wartime speeches and post-war writings— to describe the cultural, economic, military, political and intelligence ties linking their two countries.
In the decades since World War II, the notion of ‘special-ness’ has not gained much traction among scholars. This probably owes as much to the term’s conceptual imprecision as it does to the tendency of some to equate “Churchill’s compelling rhetoric...[with] the sober language of historical description” (Reynolds 1981, 166-
167). Nevertheless, certain aspects of the Anglo-American relationship were indeed unique among alliances of the World War II era. Perhaps the most distinct of these aspects was the “unusual number of contacts between officials, soldiers and ordinary citizens” from both countries (Ibid., 244).
These wartime contacts grew especially deep in the intelligence sphere, where elected leaders, top government officials, and military and civilian representatives participated in numerous informational exchanges, technical discussions and combined intelligence activities. Some of these interactions evolved into fully
162
integrated intelligence operations, as revealed in this personal reflection from a US non-commissioned officer detailed to Hut 6 at Bletchley Park:
During the entire period of operations Americans worked alongside British personnel, carrying out the same tasks under the same direction. There were no purely American sections. … There was not merely inter-allied cooperation but complete fusion of effort.171
Anglo-American intelligence cooperation persisted beyond the war as the two countries entered into more binding arrangements, beginning with the 1946 British-
U.S. COMINT Agreement (also known as the UKUSA Agreement).172 Facing the prospect of a new Soviet threat on the immediate horizon, neither party was willing to forfeit the gains of the previous five years and return to the status quo ante bellum: a period of ad hoc, intermittent intelligence cooperation between strategic competitors.
The details of this history are also important, as they highlight some of the obstacles that both parties overcame on the road to deeper intelligence cooperation.
World War I: A Beginning
Following its decisive victory at the Battle of Trafalgar in 1805, Britain enjoyed nearly a century of unchallenged naval supremacy through adherence to the two-power standard.173 Nevertheless, by the early 1900s cracks had begun to surface
171 From Lt. Colonel Roy D. Johnson to Director, SSD (Sig. Section). “Technical History of the 6813th Signal Security Service.” Item 2943. Box 970. NACP RG 457. October 20, 1945, 11. 172 UKUSA helped to put Anglo-American intelligence cooperation on a more permanent footing by defining the terms of “exchange of foreign communications-related intelligence” and associated “methods and techniques” between “the United States, the British Empire, and the British Dominions” (“British-U.S. Communication Intelligence Agreement.” TNA HW 80/4. March 5, 1945, 4). 173 The two-power standard, which Britain had pursued informally for much of the 19th century, was formally codified with the passage of the 1889 Naval Defense Act. The legislation “mandated that the Royal Navy should always possess more modern battleships and battle-cruisers than the combined forces of the next two strongest naval powers” (Seligmann 2010, 38).
163
in the Pax Britannica. Reports of an accelerated German shipbuilding effort gave rise to fevered public speculation that Berlin, buoyed by its growing economy and industrial base, would soon overtake London in its possession of modern battleships.
These concerns were fueled in part by reports from Britain’s Naval Intelligence
Department (NID) that “ships of the 1909 German programme had been ordered in advance, with materials being collected and keels being laid months before the designated time.” Top British officials deduced from these reports that Germany was rapidly and covertly accelerating its ship production schedules in order to overtake
Britain (Seligmann 2010, 38-56).174
Contrary to expectations, Berlin did not seize the mantle of naval supremacy.
Nevertheless, public concern over Germany’s great power ambitions and intentions vis-à-vis the UK continued to grow. This was seen not only in increasingly alarmist
British news bulletins, but also in the growing popularity of the German “invasion- scare” literature.175 In an effort to neutralize these perceived plots and gain greater
174 In an effort to head off these efforts and preserve its two-power standard, London responded by launching an accelerated construction program of its own—thus triggering one of the first naval arms races of the 20th century. 175 One of the first known German “invasion-scare” novels was Lt. Col. George Chesney’s The Battle of Dorking: Reminiscences of a Volunteer, first published in 1871. While the invasion literature included additional would-be conquerors (principally France and Russia), Germany was its primary focus during the early 20th century, with Ernest Childers’ Riddle of the Sands (1903) being among the most popular and influential during this period of British history. In an analysis of over 100 of these novels published in Britain from 1871 to 1914, Matin identifies 10 representative motifs: (1) near-future settings…for hostilities or narrowly averted hostilities…and corresponding displays of the wisdom of hindsight; (2) demonstrations of the vulnerability of the territories of the British Empire in addition to that of England or Great Britain; (3) depictions of invading or occupying troops and/or subversive foreigners…on British soil; (4) blendings of documented fact with fiction… (5) denunciations of incompetent (usually Liberal) British politicians…(6) governmental and public underestimations of the capabilities and nefariousness of Britain’s enemies; (7) appeals to the reader’s senses of patriotism and shame; (8) failures to support strategic and tactical innovations…; (9) geographical specificity and depictions of familiar local detail, such as English national landmarks… (10) decisive conclusions to events with Britain either defeated and the British Empire dismembered or the
164
insight into the activities and aspirations of the German Navy, the British
Government created the Secret Service Bureau (later spun off into MI5 and MI6) in
1909.
The creation of the SSB signaled Britain’s movement toward a professionalized intelligence system composed of specialized agencies. This marked a departure from the largely informal and personalized intelligence system that had existed up to this point (Ferris 2008, 532-533). This UK intelligence overhaul would reap significant dividends during World War I, when London leveraged “the most sophisticated and wide-ranging intelligence assessment activities ever done to that date” to effect a global “strangulation of the enemy industrial mobilization capacity,” as well as influence the foreign policies of neutral parties that it hoped to coax over to its side (Kennedy 2007, 699 and 716).
Central among these neutral parties was the United States, a rising economic and naval power in its own right. Similar to Britain during the 19th century, the
United States had leveraged intelligence to varying degrees, albeit never in any systematic or professionalized fashion. That remained true well into the early 20th century, a period during which it did not experience any galvanizing episodes like the
German Naval Scare that might have provided the impetus for systemic intelligence reform (Spence 2004, 515).
US intelligence capabilities during this period were comparatively limited and weak. Near the end of Theodore Roosevelt’s presidency, for instance, US foreign intelligence gathering efforts reportedly consisted of a handful of “naval and military
nation and empire strengthened by the experience of actual or threatened invasion (Matin 2011, 804).
165
attaches collecting mostly unclassified information on other armed services, supplemented by the spasmodic use of frequently unreliable part-time agents”
(Andrew 1995, 29). US counter-espionage capabilities were similarly immature.
During the early years of World War I, for example, foreign intelligence services regularly exploited America’s inability to track and monitor foreign operatives on US soil (Spence 2004, 512).
Some of the first documented Anglo-American intelligence encounters can be traced back to this period, when Britain began to increase its intelligence presence in the United States while at the same time engaging in limited cooperation with US law enforcement and intelligence agencies. Britain had several motivations for doing this.
Chief among these were a need to assure the continued flow of US-manufactured goods and capital (specifically, shipments of munitions, raw materials and foodstuffs threatened by German sabotage and U-boat attacks), neutralize the outbound flow of enemy contraband and counter German spy recruitment activities in the United States
(Warner 2004, 77; Spence 2004, 515-516; Landau 1937, 150). Through a combination of intelligence gathering, counter-intelligence efforts, and the application of direct pressure on prospective buyers and sellers of enemy contraband, Britain managed to maintain its North American supply line while at the same time denying
Germany the same advantage.
Britain also leveraged its intelligence services to bring the United States into the war “by fair means or foul” through an intensive campaign of anti-German influence operations (Jeffreys-Jones 2000, 9). Notable examples include the circumstances surrounding Britain’s disclosure of the Zimmermann Telegram to the
166
US Government176 as well as its aggressive disinformation campaign waged in the pages of American national media outlets such as the Providence Journal and the
New York Times.177 By focusing the attention of the US Government and the
American public on anecdotes and actions that painted Germany in an unfavorable light, British Intelligence helped to tilt US foreign policy in favor of the Allies and against the Central Powers (Kahn 1999, 147; Spence 2004, 522-523; Warner 2004,
78).
America’s official entry into the war in 1917 as an associate power ushered in a new phase of Anglo-American intelligence relations. Britain began to look upon the United States as not merely a useful target of intelligence collection and influence operations, but also a potential intelligence partner. Section V, a heretofore secretive
New York outpost of the Secret Intelligence Service (SIS/MI6), came out of the shadows to cultivate working relationships with the NYPD’s Intelligence branch, US
Naval Intelligence and Secret Service to counter German-backed Indian and Irish
176 Following the Admiralty’s interception and decryption of the Zimmermann telegram, the British Director of Naval Intelligence immediately saw the potential impact that the document could have on US neutrality. Captain Reginald ‘Blinker’ Hall recognized that if anything could persuade Washington to rally to the Allied cause, it would be the revelation of a secret German proposal to form an alliance with Mexico against the United States. Before this information could be shared with the United States, however, Hall had to find a way to mask the manner in which Britain had acquired the information (read: covert monitoring of US diplomatic cables, on which Germany had sent the encoded message). After developing a plausible cover, Hall shared the encoded message, solution key and its solution with the US Embassy in London. Convinced of the message’s authenticity, the US Embassy forwarded it directly to President Wilson. By early March the story had become first page material for US newspapers, helping to cement anti-German feeling in the United States and leading to a congressional declaration of war the following month (Kahn 1999; Von Zur Gathen 2007). 177 The New York-based British Naval Attaché, Guy Gaunt, famously fed false information about German espionage and sabotage efforts within the United States to John R. Rathom, the Australian-born editor of the Providence Journal. Rathom published many of these stories in his paper as part of a series of sensational exclusives that were reprinted nationwide by other newspapers. In 1918 the Justice Department silenced Rathom by securing a confession from him for his role in “fabricating sources that did not actually exist.” While it is difficult to assess with any certainty the overall impact of Rathom’s exposes on US attitudes toward Germany, historians generally cite the Providence Journal articles as emblematic of Britain’s anti-German disinformation campaign during this period (MacDonnell 1995, 20; Landau 1937, 150).
167
separatist movements operating on US soil. Section V also developed high-level ties with the White House during this period (Spence 2004, 521-523).
British military intelligence organizations also began to engage in limited information sharing with their US counterparts. Notable in this respect was Room
40’s sharing of operational intelligence with the US Navy, whose HUMINT-centric
Office of Naval Intelligence (ONI) proved insufficiently equipped to satisfy the US military’s growing demand for this information. 178 The United States reciprocated with the creation of a Joint Information Council to facilitate the sharing of naval, military, scientific, technical, and industrial research information among the Allied
Powers (Soybel 2005, 12-14).
Both countries nevertheless maintained significant reservations about taking their relationship any further. The US Navy, for one, was internally conflicted over whether to pursue deeper intelligence cooperation with the British Admiralty. Some officials, such as Rear Admiral William Sowden Sims, favored closer intelligence ties. Others, such as Chief of Naval Operations William Shepherd Benson and
Captain Roger Welles, were leery of tethering the fortunes of US Naval intelligence to those of the Admiralty, which they expected to remain as America’s principal strategic competitor in the post-war period (Ibid.).
Similar reservations were evident within the British Intelligence establishment. Sir William Reginald “Blinker” Hall, the British Director of Naval
Intelligence (DNI) and chief of the aforementioned Room 40, preferred to share information with foreign governments only when he was assured of receiving items
178 Room 40 was the name of the Royal Navy’s legendary codebreaking operation. It was run out of Room #40 in Britain’s Admiralty Ripley building under the auspices of the Naval Intelligence Division (NID), led by Captain (and later Admiral) Reginald “Blinker” Hall (Larsen 2010, 685-686).
168
of greater or equal value in return. Given the relative dearth of intelligence supplied by the United States to Britain during World War I, it is not surprising that Hall’s department was highly selective in what it was willing to share with Washington on
German U-Boat attacks. By war’s end, Britain had little incentive to pursue deeper intelligence ties with the United States. As Soybel notes, “[r]emaining constant from
World War I was the British belief that there was not much they could gain from the
Americans when it came to the gathering and analysis of…intelligence” (Ibid., 14-
15).
* * * * * * * * * * *
In the absence of a common military adversary, the United States grew less tolerant of Britain’s postwar intelligence presence within the country, as well as its perceived efforts to manipulate US foreign policy. US Government agencies complained of British intelligence officers forcing their way into Espionage Act- related interrogations, pressing for access to official reports on Irish nationalist groups, and ramping up domestic operations against US-based Communist organizations. The matter came to a head in January of 1920, when the State
Department formally called on London to terminate its intelligence gathering efforts on US soil. Soon thereafter, Britain shuttered Section V, scaled back its US intelligence presence and quietly relocated to the New York Consulate (Spence 2004,
530-533). Over the next two decades intelligence liaison between the two countries came to an effective halt.
169
The road to the special relationship
It was not until the late 1930s, when the prospect of war with Germany and
Japan loomed on the immediate horizon, that Washington and London began to consider reactivating their military and intelligence ties. Britain took the lead. In an influential policy paper published in July of 1940, the Foreign Office (FO) argued,
“the future of our widely scattered Empire is likely to depend on the evolution of an effective and enduring collaboration between ourselves and the United States." To jump-start this collaboration, the FO recommended that Britain relax its quid pro quo standard by making “spontaneous offers of co-operation” to the United States
(Reynolds 1981, 117-122).
London moved quickly to implement its new policy in the intelligence sphere.
After failing in June to persuade the US Naval Attaché to provide information on
Japanese ciphers, the British Government succeeded in convincing President
Roosevelt to approve its request for “a general exchange of technical information” in
“the radio field.” (Benson 1977, 1). Roosevelt was amenable to the request and asked the Army and Naval Departments to work out the details. While the two service branches had different views on the desirability of this new policy, 179 the Army ultimately convinced the Navy to endorse its proposal for “a general Comint exchange with the British” limited initially to cryptanalytic information (and therefore excluding information about U.S. codes, ciphers and cryptographic methods) (Ibid.,
2-3; Gladwin 1999, 122).
179 As Benson notes, the US Navy—perhaps mindful of its previous contentious dealings with Room 40 and the Admiralty more generally—was considerably less keen than the Army to pursue cryptanalytic cooperation with the British (Benson 1977, 1).
170
Roosevelt approved the Army proposal in October. Bilateral intelligence cooperation commenced in earnest in January of 1941, when the joint US cryptologic mission led by Captain Abraham Sinkov180 arrived at Bletchley Park and the inaugural members of the British Joint Staff Mission arrived in Washington, D.C..
By summer’s end, Britain’s General Code & Cipher School (GCCS) had established a regular channel of COMINT exchange with the US Navy’s Op-20-G and had attached its first permanent liaison officer to Arlington Hall, home of the US Army’s Signal
Intelligence Service (SIS), later to be renamed as the Signal Security Agency (SSA)
(Benson 1977, 3-4).
Description of intelligence cooperation machinery
What emerged over the ensuing five years was a complex and complementary set of formal agreements between GCCS and its functional equivalents in the US
Army and Navy. Two of these agreements warrant special mention for their roles in shaping the machinery of Anglo-American SIGINT cooperation:
• The October 1942 Holden Agreement (Travis-Wenger)181 defined the initial basis of Naval SIGINT cooperation between the United States and Britain. It designated Op-20-G as the senior partner in the combined Allied effort against Japanese communications and called for a significant reduction of GCCS’s involvement in this sphere. The agreement also outlined the conditions under which Op-20-G would send Japanese Naval intercept and cryptographic materials to GCCS.
180 The Sinkov Mission consisted of four junior cryptologic officers from SIS (Arlington Hall) and Op-20-G. During their two-month visit, the men gained an in-depth view into the inner-workings of GCCS and several of its intercept and radio direction finding stations. The Sinkov party also received information on certain Axis and neutral cryptographic systems, as well as current GCCS exploitation efforts conducted against these countries’ commercial and meteorological codes. The two sides also shared technical equipment, including two “Purple” (Japanese cipher) machines supplied by the Americans. But as Benson writes, arguably the most significant secret shared during this visit was Britain’s revelation to their American visitors that they had solved and were currently exploiting “the German Enigma cipher machine, used by all the German armed forces” (Benson 1977, 7; Maj. Sinkov. “Report of Cryptographic Mission.” Item 3873. Box 1296. NACP RG 457. February 1941.). 181 The full text of Travis-Wenger appears in Appendix III.
171
Finally, the agreement called for “full collaboration” between the two agencies on the “German submarine and naval cryptanalysis problems.” (Erskine 1999, 192-195).
• The May 1943 BRUSA Agreement182 defined the basis of SIGINT cooperation between GCCS and the US War Department. In addition to calling for the “complete interchange of technical data and special intelligence” concerning the Axis military and air forces (including secret services), BRUSA established a division of effort in which the US assumed primary responsibility for Japanese military and air traffic and the British “assume[d] a like responsibility for German and Italian military and air traffic.” BRUSA also defined procedures governing the mutual exchange of liaison officers, dissemination of SIGINT and SIGINT-derived material, training of US SIGINT personnel by GCCS, and secure handling of cryptographic material.183
During the summer and fall months of 1943, many of the liaison provisions outlined in BRUSA began to fall into place for the Army SIGINT elements in both countries. In Britain, over 250 regular US Army personnel were integrated into the
SIGINT operations of GCCS as part of Operation BEECHNUT. After receiving on- the-job training in intercept operations, machine processing and cryptanalysis, many of these US personnel were absorbed into Bletchley Park, where they became “fully proficient in British high-grade cipher-breaking techniques” and “significantly contributed to the Enigma attack in both intercept and solution” (Smith 1993, 166;
Benson 1997, 110-111).
Additionally, US Army intelligence liaison officers were attached to the key sections and branches of GCCS, where they reviewed daily decoded messages produced by the British, handled ad hoc information inquiries, and cabled unique items of immediate intelligence to Washington and US field commands around the
182 The full text of the BRUSA Agreement appears in Appendix IV. 183 Col. O.L. Nelson and Major General George Strong. “Agreement between British Government Code and Cipher School and U.S. War Department in regard to certain ‘Special Intelligence.’” Item 2751. Box 943. NACP RG 457. June 15, 1943.
172
world. The US liaison officers were also responsible for maintaining productive relations with GCCS, developing a working knowledge of the cryptanalytic methods and achievements of GCCS writ large, and producing regular reports on the internal structure, functions and activities of GCCS.184
A much smaller number of British GCCS personnel—many of them veteran cryptanalysts on loan from Bletchley Park—were sent to Washington, D.C. to form a
Signals Intelligence Liaison Unit (SLU) responsible for performing similar functions.
The SLU consisted of a senior GCCS representative and detachments deployed to each of the three primary US military SIGINT organizations: Op-20-G at the Naval
Communications Annex, the Army G -2 Special Branch at the Pentagon, and the
Army Signal Security Agency (SSA) at Arlington Hall. A senior British Liaison
Officer (BLO) was attached to each of these detachments and reported to the Senior
GCCS representative. These senior BLOs, listed in Table 9 below, were complemented by additional technical specialists and intelligence officers from
GCCS.
Some of these integrated GCCS personnel were assigned to work on specific cryptanalytic topics of interest and produce periodic reports on their involvement in these projects. These reports were frequently shared both with the American heads of their assigned organizations and GCCS leadership in Britain.
184 Maj. Louis T. Stone, Jr. “Memorandum Describing American Liaison.” Item 9002. SRH 153. NACP RG 457. 1945; From Lt. Colonel Frank B. Rowlett to Colonel Harold G. Hayes. “Memorandum for Colonel Harold G. Hayes: Duties of the liaison officer at GCCS.” Item 2741. Box 941. NACP RG 457. April 13, 1945.
173
British Signals Intelligence Liaison Unit (SLU) Detachments in Washington, D.C. Senior GCCS Representative in United States Dates of service Capt. Edward G. Hastings May 1942 – July 1943 Col. H. M. O’Connor July 1943 – July 1945 Group Capt. Eric Malcolm Jones July 1945 – End of War
Senior GCCS Representative to SIS/SSA Maj. G.G. Stevens Dec. 1943 – Sept. 1944 Maj. J. R. Cheadle Oct. 1944 – End of War
Senior GCCS Representative to G-2 Special Branch Lt. Col. Godley Feb. 1944 – End of War
Senior GCCS Representative to Op-20-G Mr. H.R. Foss Feb. 1944 – July 1944 Mr. Bodsworth July 1944 – End of War
US SIGINT Organizations – Washington, D.C. Chiefs of US Army Signals Intelligence Service/Signal Security Agency Dates of service (SIS/SSA; Arlington Hall) Col. W. Preston Corderman July 1943 – End of War
Chief of G-2 Military Intelligence Service (MIS) Special Branch (Special Branch; Pentagon) Brig. Gen. Carter Clarke May 1942 – June 1944
Chiefs of the Office of Chief Of Naval Operations, 20th Division of the Office of Naval
Communications, G Section (Op-20-G; Naval Communications Annex) Cdr. John R. Redman Feb. 1942 – Oct .1942 Capt. Earl E. Stone Oct. 1942 – Mar. 1944 Capt. Phillip R. Kinney April 1944 – Nov. 1944 Cdr. Joseph N. Wenger Nov. 1944 – End of War
Sources: HW 57/2; HW 57/5; Benson 1997 Table 9. Anglo-American Liaison Structure and Key Personnel, 1942-1945
Given the constant bureaucratic reshuffling that occurred within the US SIGINT agencies during this period, the SLU gave its personnel relatively broad issue portfolios (e.g., topics such as ‘Japanese Army,’ ‘Japanese Air’ and ‘Japanese
174
Meteorological’), rather than assigning them to specific sections. This flexible arrangement, depicted in Figure 12, enabled the BLOs to work with the appropriate
US personnel for their assigned portfolio, regardless of their location within the evolving US SIGINT bureaucracy. It was therefore not unusual for the BLOs to visit and interact with many different sections and branches of their assigned host organization. As British Lt. Col. Godley wrote in one of his monthly reports, such flexibility was “necessary since in no other way [could] I follow ULTRA185 through all its processes here and obtain all the results of the various sections working on it.”186
Compared to 30 Mission, the members of the Washington-based SLU had significantly greater access to their hosts and their intelligence activities.
Nevertheless, it would be inaccurate to suggest that the Anglo-American intelligence relationship was devoid of complications and inefficiencies. Britain’s official unpublished “History of Liaison with Op-20-G,” for instance, contains numerous examples of the Navy’s unwillingness to share certain intelligence publications with
GCCS, as well as perceptions of a highly bureaucratic, anti-British mindset taking hold among the US Naval leadership.187
185 ULTRA was the name assigned by British military intelligence to “highly secret information obtained by cryptographic means from enemy high grade ciphers.” Per the 1943 BRUSA Agreement, the ULTRA designation was to be reserved exclusively for material that met the above definition and “emanate[ed] from the United Kingdom and [was] transmitted to Commands abroad” (Sims 1997, 35-36). 186 Lt. Colonel C. G. A. Godley. “Japanese Military Section Notes.” Monthly Letter Series: June 1944. TNA HW 57/2. July 1944. 187 “History of Liaison with Op-20-G (Washington) as carried out by Representatives of Naval Section, G.C. & C.S.” TNA HW 8/49. 1946 (Hereafter “History of Liaison with Op-20-G in Washington”).
175
Figure 12. Organizational Division of Effort for British SLU in the USA
Hypothesis Testing
Rationale for data selection
Similar to the Anglo-Soviet case study, there is no single dataset that could possibly encompass the full array of Anglo-American intelligence exchange channels that existed during World War II. In addition to the SIGINT channel described above, intelligence cooperation also occurred between American and British domestic law enforcement agencies (e.g., the FBI-BSC channel) and special operations organizations (e.g., the OSS-SOE channel), as well at the level of the Combined Joint
Chiefs of Staff via the Combined Intelligence Committee. Different types of
176
intelligence goods were exchanged within these channels, and the quantity and quality of primary source documentation varied between them.188
I selected the monthly off-the-record notes and accompanying documentation sent from Britain’s Washington-based SLU to the Director of GCCS, as captured in record sets HW 57/2 and HW 57/5 in the UK National Archives. I selected this dataset for several reasons. First, the express purpose of these notes was to provide a continuous view into the “general tendencies and off the record information” of the
SLU’s interactions with Arlington Hall, the G-2 Special Branch and Op-20-G during the post-BRUSA period, circa 1943-1945.189 I felt this type of information would be conducive to my goals of testing the observable implications of the dyadic democratic peace and relational contracting hypotheses.
Second, the monthly SLU notes provide a degree of structural symmetry with the War Diaries of the previous chapter. Both are written from the perspective of
British military intelligence liaison officers operating in an allied country during
World War II. Third and finally, the notes also serve as a source of fresh primary source material for testing the democratic peace and relational contracting hypotheses. Most of the historical literature concerning Anglo-American SIGINT cooperation during World War II has tended to focus on the American presence at
Bletchley Park. By contrast, relatively little has been written about the British
SIGINT presence at the Naval Department, the Pentagon and Arlington Hall.
188 Bailey 2000. KV 4/447; V. Dykes and W.B. Smith. “Directive by the Combined Chiefs of Staff for Combined Intelligence.” TNA CAB/122. February 11, 1942. 189 From W. Bodsworth to Director, GCCS. “Op-20-G Notes for July 1945.” TNA HW 57/5. July 1945, 4.
177
Testing the dyadic democratic peace argument
Measuring the dependent variable
Frequency of contact. As seen in the previous chapter, frequency of contact can be a useful proxy for measuring the depth of intelligence cooperation in highly ad hoc and transactional relationships such as the Anglo-Soviet WCID. The members of
30 Mission were physically walled off from their Soviet counterparts and the vast majority of their interactions were tightly regulated by the Otdel. Against this backdrop, all face-to-face meetings were ipso facto noteworthy events that warranted explicit mention in the 30 Mission War Diaries.
Frequency of contact is a more problematic proxy for measuring the depth of intelligence cooperation in continuous and tightly integrated relationships, such as the
Anglo-American WCID. When both intelligence partners are working directly with one another on a continuous basis, the participants have less incentive to record each face-to-face exchange as a discrete event. This was certainly the case for the British
SLU, whose personnel were far more focused on reporting the general state of affairs within their host organizations and conveying the status of ongoing joint cryptologic efforts.
This is not to suggest that the monthly SLU notes contain no evidence of direct, face-to-face interactions between British and American intelligence personnel.
Figure 13, for example, illustrates the monthly ebb and flow of British visitors to Op-
20-G, G-2 Special Branch and the SSA. An important point to consider when viewing this figure is that a vast majority of the recorded events were non-recurring.
In fact, many of them involved visiting delegations of British personnel from
178
Bletchley Park. Figure 13 is thus an imperfect measurement of contact frequency because it under-represents the total number of face-to-face encounters that occurred between the SLU BLOs and their American counterparts.
Figure 13. Anglo-American Intelligence Exchanges (Washington), 1943-1945
Two additional aspects of Figure 13 are worth noting. The first is the upward trajectory of encounters from December of 1944 through March of 1945. This is perhaps best explained as a function of an increased Allied need for detailed intelligence on Japanese commercial shipping during the final months of the war.190
Indeed, many of the visits organized during this period relate in some way to joint
US-British efforts to exploit Japanese commercial and shipping ciphers. A second notable aspect of Figure 13 is the pronounced decline of encounters from April of
190 See, for example: From Godley to Director, GCCS. “Japanese Military Section Notes for January 1945.” HW 57/5. February 1945, 3; From Maj. Cheadle to Director, GCCS. “Arlington Hall Notes for February 1945.” HW 57/5. March 1945, 6.
179
1945 through war’s end. The potential causes of this decline are less evident. One possibility is that there was simply a diminishing need for UK personnel to visit US
SIGINT installations as the Japan campaign drew to a close. Another is the degree of negative fallout resulting from Britain’s under-classification of an Admiralty document containing sensitive information derived from material shared by the
United States. As discussed later in this chapter, this security violation sparked a minor crisis in Anglo-American intelligence relations because it resulted in the disclosure of sensitive information concerning Anglo-American SIGINT in non-
Special Intelligence (SI) channels.
While Figure 13 offers a partial view into the ebb and flow of Anglo-
American wartime intelligence cooperation, it is potentially misleading because it does not reflect the numerous daily interactions between the permanently attached
GCCS personnel and their co-located American colleagues. As one participant made clear, the period of late 1943 through 1945 was characterized by “direct communication between working members of all the cryptographic sections and subsections in Washington and their opposite numbers in G.C. & C.S.”191 Also missing from Figure 13 are the numerous inter-departmental exchanges that occurred during the war, wherein GCCS liaison officers attached to one US military SIGINT agency (e.g., Arlington Hall) met with US personnel stationed at another (e.g., Op-20-
G).192
191 “History of Liaison with Op-20-G in Washington.” TNA HW 8/49. 1946, 25-26. 192 I noted at least two cases of this occurring in June of 1945. It was unclear in each instance how long the inter-departmental liaison arrangements had been in place, as well as how long they persisted. Specifically: • Maj. J.R. Cheadle, a permanently attached GCCS Liaison officer to SSA (Arlington Hall) is reported to have “visit[ed] Op-20-G about three times a week, keeping himself abreast
180
Table 10. SLU Presence at Select US Military SIGINT Agencies, 1944-1945
In an effort to compensate for the limitations of the ‘frequency of contact’ proxy for this case, I devised a complementary heuristic to measure the extent to which members of the SLU interacted on a continuous basis with their US SIGINT counterparts. This heuristic, displayed in Table 10, conveys whether a BLO filed at least one off-the-record note from each US SIGINT agency from February of 1944 through June of 1945. What emerges in this table is a highly stable and continuous
SLU presence throughout the study period.193
of the cryptographic problems and maintaining useful contacts…” (From Cheadle to Director, GCCS. “Arlington Hall Notes for June 1945.” TNA HW 57/5. July 1945, 1). • Lt. Cdr. C.J.B. Chalkley, a permanently attached GCCS Liaison Officer to Op-20-G (Naval Department), is reported to have visited the Economic Branch of MIS, also known as the G-2 Special Branch, approximately twice per week (From Lt. Cdr. Chalkley to Director, GCCS. “Op-20-G Notes for June 1945.” TNA HW 57/5. July 1945, 6). 193 The lone gap in this graphic—a three-month period during which no SLU notes were filed for the topic of Japanese Air—is explained by a bureaucratic re-organization within Army Intelligence (G-2) that took several months to complete.
181
Taken together, Figure 13 and Table 10 provide evidence of a relatively high and stable frequency of Anglo-American intelligence cooperation from late 1943 through war’s end. Compared to similar measurements recorded for the Anglo-Soviet
WCID over the same time period (as seen specifically Figures 8 and 9), the Anglo-
American WCID is in certain respects a polar opposite. While 30 Mission was facing a steady decline in the number of face-to-face contacts with the Soviet General Staff, the SLU was at best gaining momentum (as seen in the rising number of monthly visits from late 1944 through March of 1945) and, at worst, maintaining a fully integrated presence in each of the primary US SIGINT organizations.
Nature and granularity of intelligence cooperation. As documented in the previous chapter, 30 Mission and the Soviet General Staff exchanged intelligence goods of varying degrees of granularity. By contrast, cooperation between Britain’s
GCCS and America’s Op-20-G, the G-2 Special Branch and SSA was focused almost exclusively on SIGINT. Given that SIGINT rates as one of the most granular intelligence goods (per Table 3)194 and that the ‘Special Intelligence’ exchanged between GCCS and its US counterparts was of a higher grade than the limited amount of ‘Y’ material (e.g., German police communications) exchanged between 30 Mission and SGS, it is reasonable to conclude that the Anglo-American WCID exhibited consistently higher degrees of granularity than the Anglo-Soviet WCID.
Measuring the independent variable
Consistent with the approach taken in the previous chapter, I measured the value of the independent variable and determined whether its observed values aligned
194 SIGINT registers a granularity score of 6—the highest rating of any intelligence good listed in Table 4, tied with Targeting, Captured Documents, Enemy Intelligence, POWs: Allied/Evacuation, and Chemical Warfare.
182
with the predictions of the dyadic democratic peace argument. I did this once again by leveraging the polity2 values from the Polity IV dataset for the United Kingdom and the United States from 1943 through 1945. As expected, the polity values are very high for both countries. Each registers a maximum score of 10 (Marshall,
Jaggers et al 2011). This places the Anglo-American WCID in the northeastern quadrant of the property space diagram depicted in Figure 6, thus making it an ‘ideal type’ fully democratic WCID.
The presence of two fully institutionalized democracies and high observed levels of intelligence cooperation between them are congruent with the dyadic democratic peace hypothesis. As noted previously, however, evidence of congruence tells us nothing about how or why intelligence cooperation flourished to the extent that it did between Britain and the United States. To gain these insights we must once again explore the causal logic of the dyadic democratic peace argument, looking specifically for evidence of the regime recognition dynamic and its observable implications within this case.
Testing for observable implications
A key insight of the dyadic democratic peace proposition, as adapted to this study, is the idea that fully democratic intelligence services will behave differently toward one another over time than they will toward autocratic intelligence services.
Consistent with this regime recognition dynamic, members of an intra-democratic
WCID are expected to benefit from a mutual presumption of amity because they both operate from a menu of shared understandings comprising contingent consent and
183
bounded uncertainty.195 That presumption is expected to compel each party to eschew coercive tactics and turn to institutionalized procedures for resolving disputes
(Dixon 1994, 16). More specifically, each party is expected to discount the possibility of defection and ‘keep the game going’ to one another’s benefit—even during periods of disagreement. This dynamic is expected to lead to comparatively higher levels of bilateral intelligence cooperation over time.
Observable implications of the dyadic democratic peace proposition (1) Britain will respect the outcomes of agreements favoring the United States, and the United States will not obstruct Britain Contingent from pursuing greater influence within the Anglo-American consent intelligence relationship. predictions (2) Where conflicts or disputes arise with the United States, Britain will seek to resolve them through mutual accommodation.196 (3) The United States will meet Britain’s minimum standards of expected behavior as an intelligence partner: a) British personnel will be afforded direct, working-level access to their American counterparts. Bounded b) Informational requests on matters concerning the common uncertainty enemy will be answered in a prompt and satisfactory predictions fashion. c) The US will afford British personnel sufficient latitude to carry out their official liaison duties at their assigned organizations. Table 11. Dyadic Democratic Peace Argument: Anglo-American Predictions
If this application of the dyadic democratic peace argument is to find empirical support, we should once again see evidence of its causal mechanism, the
195 Earlier in Chapter Two, contingent consent was likened to a gentleman’s agreement, wherein each party agrees to respect transactional outcomes favoring the other side provided that the ‘winner’ in a given transaction agrees not to exploit its temporary advantage over the ‘loser’ in subsequent interactions. Bounded uncertainty was defined as a “predictable and mutually acceptable range” of behaviors constituting fair play between the parties (Dixon 1994, 15-16). 196 In this case, peaceful accommodation refers to what Dixon describes as “any written or unwritten mutually agreeable arrangements between disputant parties that at least temporarily resolve or remove from contention one or more—but not necessarily all—of the issues of the underlying dispute” (Ibid., 18).
184
regime recognition dynamic, at work within the case. Table 11 lists two sets of propositions, each of which addresses a specific causal mechanism of the argument.
Case prediction #1: Britain will respect the outcomes of agreements favoring the United States, and the United States will not obstruct Britain from pursuing greater influence within the Anglo-American intelligence relationship.
As Britain and the United States developed closer intelligence ties, each side had to make sacrifices and cede some of its traditional intelligence prerogatives in the interest of defeating the common enemy. Neither could continue to justify the luxury of maintaining unilateral intelligence capabilities strictly for their own sake. What emerged instead was an intelligence division of effort based upon the principle of comparative advantage. Each side focused on the intelligence problems and targets that it was best positioned and most well-equipped to address.
This division of effort was codified in understandings such as the aforementioned Travis-Wenger Agreement of 1942, which formally recognized the
United States and Britain as the “coordinating heads” for naval SIGINT in the Pacific and Atlantic theaters, respectively. Travis-Wenger also clarified the implications of this division of effort, as seen in its call for “the British [to] withdraw from active cryptanalytical work in the Pacific Area” and the Americans to assume “the general direction and control of the effort against Japanese communications” (Erskine 1999,
194). In addition to suspending most of its active SIGINT efforts against Japan,
GCCS agreed to “abandon naval cryptanalysis at [its] Kilindini” station in East Africa and “disband the British-Australian naval unit at Melbourne and turn over to the U.S. unit there such personnel as the U.S. may desire” (Ibid.).
185
From Britain’s perspective, these were not trivial sacrifices. It is worth recalling that, in a relatively short period during the 1930s, the GCCS Naval Section had developed a global SIGINT collection posture against numerous foreign powers, to include Japan and the Soviet Union (McKay 1997, 4). The prospect of scaling back the scope of these activities to accommodate the United States—one of its principal and historic naval rivals—did not set well with the British Admiralty and
GCCS leadership (Smith 1993, 127). Nevertheless, Britain’s thinking on this began to change in 1942, when a string of Japanese military advances forced London to shutter and relocate its Japan-focused SIGINT units on three successive occasions.
At the same time as London was scaling back its global SIGINT collection posture, the United States was beginning to ramp up its own. In the months following the Pearl Harbor attacks, the Navy’s Op-20-G had begun to overtake GCCS in its collection of Japanese naval communications and exploitation of Japanese naval codes. Meanwhile, GCCS found itself without a sufficient number of linguists and technical personnel to perform Japanese naval cryptanalysis (Erskine 2006, 2).
Against this backdrop, Japanese cryptographic exchange was fast becoming a “one – way-only” affair between the United States and Britain, with the former supplying nearly all of the raw material and taking the lead in cryptanalytic work.197 GCCS and the Admiralty thus had little choice but to accede to the terms of Travis-Wenger.
While these developments led some senior British military leaders to conclude that Britain had lost all leverage in its larger strategic relationship with the United
States, the BLOs assigned to the SLU did not see it that way. The monthly off-the-
197 History of Liaison with Op-20-G in Washington. TNA HW 8/49. 1946, 6.
186
record SLU notes contain numerous examples of BLOs seeking to exert influence and advocate British positions within their assigned US host organizations. Consistent with the dyadic democratic peace argument, the US SIGINT agencies proved generally willing to tolerate—and in certain cases, act upon—these recommendations in ways that fundamentally reshaped the Anglo-American intelligence relationship, as well as the structure and function of the agencies themselves.
One notable example of this is seen in Lt. Col. Lewis’s efforts to compel
Arlington Hall to reconsider its highly stratified organizational setup, in which individual sections (Cryptanalysis, Cover Control and Traffic Analysis) worked in isolation from one another without considering the collective implications of their activities. As Lewis observed in one of his monthly SLU notes, the Arlington Hall leadership was not only willing to listen to his recommendations, but also seemed amenable to implementing some of them:
Last month I mentioned that there was a real need for a co-ordination or fusion process here and for some sort of combined directive for cover. … I brought up the question with [S.S.A. Chief, Col. W. Preston] Corderman and [Col. Harold G.] Hayes and they agreed that there was a need for a regrouping of effort. My main points were that there was room for closer working between the three main branches (Crypt, Cover control and T.A.), that a system of individual liaison officers between one section and another was not sufficient and that the whole co-ordination should be built on an area basis. … After an initial discussion of the proposal it was decided to call a meeting to discuss the suggestions. This took place on July 24th and was attended by Corderman, Friedman, O.C. B Branch, E. Branch, B II, B IV, and a dozen or so other representatives. I was asked to put forward the proposals and after doing so asked for comments. I was prepared for some dissent, but fortunately further arguments were not necessary as the scheme was accepted by the section heads. … I was then asked if I would write up the scheme and circulate it to the parties concerned. The proposals have been put forward in draft form and … Hayes…has asked me if I would suggest by name the members of the fusion party.198
198 From Lt. Col. Lewis to Director, GCCS. “Report on Attachment to S.S.A.” TNA HW 57/2. July 1944, 1-2.
187
Lewis’s recommendations would prove more difficult to implement than he had initially anticipated. He encountered resistance from SSA section officers who were skeptical of the value added by this new approach, as well as from American traffic analysts who were reluctant to incorporate new information sources to which they were unaccustomed. Nevertheless, by gaining and maintaining the attention of senior SSA leadership, Lewis ultimately helped to shape a more comprehensive and fused US intelligence effort against Japanese naval communications.
The SLU exerted influence in more subtle ways as well. This is seen in the efforts of BLOs to introduce and acquaint themselves with American SIGINT officers
“who [could prove] useful to [GCCS] for various reasons”—both through interacting with them “in the ordinary course of work” and “meet[ing] them socially to get to know them well.”199 The US SIGINT agencies did very little to discourage these
British attempts to curry influence. This is seen indirectly in an observation by
Britain’s senior SIGINT representative at Op-20-G, Mr. William Bodsworth, to the
Director of GCCS:
Your remark about the expansion of [US Navy Lt. Cmdr. G.C.] Manson’s [Atlantic Cryptanalysis] party after the German way is also noted. Manson has recently acquired another room into which I have not been invited yet, but ways of getting in are not difficult to find and I shall report on it shortly. I suspect it contains his machine people (Spanish ENIGMA and Italian Republican).200
Bodsworth’s observation betrays a quiet confidence in the SLU’s ability to maneuver within the US military SIGINT establishment and gain access to certain (but by no
199 From Bodsworth to Director, GCCS. “Op-20-G Notes for January 1945.” TNA HW 57/5. January 1945, 2-3. 200 From Bodsworth to Director, GCCS. “Op-20-G Notes for February 1945.” TNA HW 57/5. March 1945, 4.
188
means all201) US intelligence resources—even those that had not yet been made available to the British. There is considerable anecdotal evidence to support
Bodsworth’s position. In one notable instance, a BLO reported that he had personally
“discovered about sixty captured documents concerning Japanese communications which had been stowed away and not translated” during a recent tour of Op-20-G’s
Collateral Section—and subsequently obtained a summary of them for a fellow
GCCS colleague.202 Such a degree of access to highly sensitive materials203 provides a marked contrast with the obstructive approach taken by the Soviet Otdel, which regularly blocked social exchanges and official meetings between the Soviet General
Staff and their Moscow-based British counterparts.
Case Prediction #2: Where conflicts or disputes arise with the United States,
Britain will seek to resolve them through mutual accommodation. Notwithstanding the overall strength of the Anglo-American intelligence relationship, there were contentious episodes that threatened at various points to derail cooperation. One of
201 As the GCCS Naval Section noted in its official history of liaison with Op-20-G, there was at times “an unwillingness [on the part of Op-20-G] to exchange material” with the Washington-based representatives of GCCS. The authors of this history specifically noted that, “the fact that several important American intelligence reports available in PSIS [Pacific Strategic Intelligence Section] were not sent to G.C. & C.S. and that many of the officers within the section were engaged on studies which the British would not receive, was a considerable handicap to cooperation” (“History of Liaison with Op-20-G in Washington.” TNA HW 8/49. 1946, 36). 202 From Chalkley to Director, GCCS. “Op-20-G Notes for January 1945.” TNA HW 57/5. February 1945, 6-7. 203 There is strong evidence to suggest that this degree of access extended beyond SIGINT channels. Christopher Andrew cites the reflections of an American OSS counterintelligence officer who was detailed to the Double Agent Section of Britain’s MI5 during World War II: For even an Ally to be admitted to a full access to all secret files and to a knowledge of their sources; to information on most secret methods and procedures; and to a knowledge of personnel and the system of organization and of operations—in short to the innermost arcana, in this case, of perhaps the world’s most experienced and efficient, and therefore most carefully safeguarded, security systems—was beyond precedent or expectation. Yet the British did it. The implications of this fact are staggering—and completely inexplicable in terms of merely cheap exchange of mutual advantage’ (Unpublished X2 History, as cited in Andrew 1995, 139-140).
189
the most well-known examples occurred during the winter months of 1942 to 1943, when the US Chief of Army Intelligence (G-2) accused GCCS of withholding cryptographic material from the United States and leveraging ‘back door’ methods to gain access for one of its most prominent cryptographers (Dr. Alan Turing) to a highly sensitive voice-scrambling project at Bell Laboratories in New Jersey. These allegations ultimately led the Chief of the British Joint Staff Mission to warn General
George Marshall of unspecified ‘unfortunate effect[s]’ that would occur, should
Turing’s visit be blocked by the Americans (Benson 1997, 97-106; Gladwin 1999,
130-132; Smith 1993, 136). A mutually acceptable resolution was eventually reached, but not until after the matter had been referred to the most senior levels of US and
British military leadership.204
To the extent that the secondary literature has addressed wartime conflicts between the British and US intelligence services, much of the focus has centered
(understandably) on a handful of dramatic episodes involving significant historical figures, such as the Bell Laboratories/Alan Turing incident described above. Less attention has been given to the far more numerous and lower-level disputes that occurred between members of the Washington-based SLU and their American counterparts at Op-20-G, Arlington Hall and the Pentagon. An examination of two of these incidents will help to establish whether, as predicted by the dyadic democratic peace argument, the British placed a greater emphasis on keeping the game going
204 The United States ultimately granted Turing authorization to visit the facility, following British assurances to General Marshall that GCCS was not withholding any of the alleged information and was moreover “prepared to share everything desired with the US Army if the Americans would come to England” (Gladwin 1999, 132).
190
with their US counterparts than in ‘keeping score’ and retaliating in response to perceived breaches of trust.
Use of the BRUSA secure communications circuit. As noted above, the 1942
Travis-Wenger Agreement outlined an explicit geographic division of labor between the Naval Section of GCCS and Op-20-G. Travis-Wenger further established that
Op-20-G, as the senior Allied partner responsible for Japan, would assume “general direction and control of the effort against Japanese communications” and supply the
British with Japanese radio intelligence, code and cipher key recoveries, and “as much Japanese intercepted raw traffic as possible.” Britain, for its part, agreed to suspend its naval cryptanalysis efforts against Japan whilst “maintain[ing] a research and intelligence unit at G.C.& C.S. so as not to lose touch with the Japanese problem.” This research unit was to be supported by Britain’s Far East Combined
Bureau (FECB), a mobile GCCS out-station that moved during the war between
Colombo, Ceylon (later Sri Lanka) and Kilindini, Kenya (Erskine 1999, 193-194).
Despite bringing unprecedented clarity to Anglo-American naval intelligence relations, Travis-Wenger was found wanting in several respects by the British. To begin with, it did not allow for the direct and efficient transmission of Japanese radio intelligence between Op-20-G and the FECB. According to the agreement, Op-20-G was to pass all such materials indirectly to the FECB via GCCS headquarters (Ibid.,
193). This introduced costly delays into bilateral intelligence cooperation. During its time in Kilindini, for example, the FECB had to wait for an average of 12 additional days to receive raw Japanese material from Op-20-G—thus ensuring that the information was often dated or obsolete by the time of its ultimate arrival (Erskine
191
2006, 2). Another problem concerned the tightly circumscribed geographic parameters of Travis-Wenger. The agreement authorized the British to receive only radio intelligence “bearing upon operations in the Indian Ocean,” which excluded much of the radio intelligence concerning the Pacific Theater (Erskine 1999, 193).
These problems, coupled with Britain’s diminished collection posture against
Japan, ultimately led GCCS to press for substantive modifications to Travis-Wenger.
Instead of relying on slow and indirect long-haul transport of Japanese raw intercept,
GCCS and the FECB preferred to receive this material directly via the US Navy’s secure radio circuit (known as both “COMB” and “TUNA”), which was utilized for the transmission of operational and non-operational intelligence. This was a non- starter the US Navy. By order of the US Chief of Staff, the circuit in question leveraged a highly sensitive cipher that was “not to be shared with the British—or even seen by them” (Erskine 2006, 6).
GCCS and the Navy ultimately reached an accommodation on this matter when they signed the January 1944 Naval BRUSA Agreement, which authorized the creation and use of a common circuit (called BRUSA) capable of reaching both
British and US Navy SIGINT Units, as well as top military commands (Ibid., 7;
Benson 1997, 120). The agreement further stipulated that OP-20-G and GCCS would utilize Combined Cipher Machines (CCMs) on the new BRUSA circuit, thus ensuring that each side could encrypt and decrypt traffic sent by the other.
In principle, both GCCS and Op-20-G were supposed to transmit all Japanese radio traffic via BRUSA. Operationally relevant traffic was to be addressed to the
“DISC” collective routing address (analogous in function to COMB) which included
192
British and US Navy units, as well as top command elements from both countries.
The “DAZE” address (analogous in function to TUNA) was to be utilized for non- operational messages, which were not transmitted to top commands.
In practice, Op-20-G quickly discovered that the BRUSA circuit could not accommodate the high volumes of traffic being sent. As a result, Op-20-G was forced to scale back its transmissions on the BRUSA circuit and revert to utilizing the
US-only COMB circuit instead (Erskine 2006, 7). This had the effect of reducing the total volume of operational intelligence received at GCCS and its outstations.
This development did not set well with GCCS, which began to vocalize its objections via the SLU with Op-20-G. This became a point of contention between the two sides, as revealed in a BLO’s observation that his recent interactions with the
Pacific Correlation and Dissemination (GI-P) section had gone “well until [he had] raised a question bearing on the DISC/COMB circuit, whereupon [Section
Commander] Daisley hedged and the conversation became slightly sticky.”205
By April the situation had worsened to the point that Colonel H.M. O’Connor, the senior GCCS representative in Washington, complained about the “rather niggardly interpretation” that the chief of GI-P was placing on his “orders for sending out decodes on DISC.” This time there was concrete evidence of a breach by the
Americans. According to O’Connor, “[t]hree signals of vital interest” had been sent by Op-20-G via COMB, but not via DISC as per the agreed-upon policy. What made this information especially credible was its source: Commander Leggatt, a sympathetic US naval officer assigned to the ONI’s SIGINT branch (F-22), who had
205 From Chalkley to Director, GCCS. “OP-20-G Notes for February 1945.” TNA HW 57/5. March 1945, 1-2.
193
seen these items himself and wondered why they had not been transmitted to the
British via DISC.206
Despite conceding that Britain could “hardly have a better case” of an
American breach of the Naval BRUSA Agreement, O’Connor had reservations about pursuing the matter with senior US Naval leadership. By his reasoning, Britain was already “getting all the stuff” it needed through other channels. There was also strong reason to believe that this would continue, given Leggatt’s pledge to the SLU that he would continue monitoring COMB and forwarding any items that were not reaching
GCCS as they should. Rather than recommending a specific course of action,
O’Connor put the decision back on the GCCS director:
What do you feel about this? I did not like to implicate Leggatt by taking this up myself with Op-20 but it is open for you to do so as soon as you see the items in the R.I. Summary. … If on the other hand you are satisfied with things as they are, you may prefer not to make an issue of NEGAT’s failure to use of [sic] BRUSA for items which you think should be on it. If you are doubtful I suggest you send me a signal of your views, referring to these items, for me to discuss with Director on his arrival here, when he can decide whether or not to take it up with Wenger.207
While allowing that Britain would be within its rights to protest perceived US violations of the Travis-Wenger and Naval BRUSA Agreements, O’Connor seemed to favor dropping the matter and ‘keeping the game going’ with his American hosts, rather than pursuing it further and risk squandering his social capital with
Commander Leggett.
The under-classification of CB 4377. As discussed in Chapter Two, the decision to cooperate with another country’s intelligence services is often predicated
206 From Col. O’Connor to Director, GCCS. “Monthly Letter Series.” TNA HW 57/5. April 7, 1945, 3. 207 Ibid.
194
upon a variety of risks and considerations. Chief among these is the partner’s perceived ability and commitment to protect sensitive information, once it has been shared (Sims 2006, 205). Protection in this case can refer to the partner’s ability to guard against leaks, its commitment to sound communications security practices, and its institution and adherence to strict ‘third-party’ rules. During World War II, both the American and British military intelligence establishments expressed periodic reservations about the other’s bona fides in this regard. This is particularly evident in the language of Travis-Wenger:
• “The primary concern of the British over U.S. entry into the German field is on the question of security. The British treat German Enigma matters on a much higher plane than any other which they handle. They believe that…any disclosures [of Enigma] might defeat the entire effort in every field. The British consider that, in going into the work, the U.S. should be prepared to accept British standards of security and insure compliance therewith” (Erskine 1999, 194).
• “As for security, the U.S. feels great concern over the treatment of intelligence obtained from Japanese naval communications and desires that it be handled on the same plane as the German.” (Ibid., 194).
These concerns came to a head in June of 1945, following the British
Admiralty’s accidental under-classification of a technical manual for UK intercept operators. The document, which came to be known by its serial number (CB 4377), had been derived in part from Top Secret SIGINT supplied by the United States.
When the US Navy discovered that CB 4377 had been published at a lower classification (British Confidential) with a much broader distribution, it immediately instituted a “general freeze order, regarding relations with the British, to all sections” at Op-20-G.208
208 From Chalkley to Director, GCCS. “OP-20-G Notes for June 1945.” TNA HW 57/5. July 1945, 1.
195
The SLU personnel at Op-20-G felt the effects of this freeze immediately.
Some reported that their American colleagues had become suddenly reluctant to work and exchange information with them. Others reported that the problem had expanded beyond Op-20-G. In one notable incident, a US Army Lieutenant at Arlington Hall inexplicably retracted his offer to sponsor the visit of a BLO.209 The BLO subsequently learned that his American hosts were keeping “a careful check…on all his activities since his arrival” and producing “a weekly report on every request he had made” of the various sections within Op-20-G.210
The British moved quickly to contain and minimize the potential fallout from this incident by imposing a week-long moratorium on “any requests from GCCS for material or information, even of a routine nature.”211 This helped to buy time for the
Washington-based BLOs to focus on the problem at hand and identify a mutually acceptable solution for the Admiralty and Navy. This was finally achieved in July, when the British agreed to upgrade the classification of CB 4377 to Top Secret. The
Americans, in response, “expressed satisfaction that the…incident was now closed and regret[ted] that [they] had to adopt the line taken.”212
Despite creating a temporary crisis in bilateral naval intelligence liaison, the under-classification of CB 4377 did not have any permanent or long-lasting effects on
Anglo-American intelligence cooperation writ large. Consistent with the causal logic of the dyadic democratic peace argument, GCCS worked with Op-20-G to remove the
209 From Lt. Laslett to Director, GCCS. “Op-20-G Notes for July 1945.” TNA HW 57/5. July 1945, 2. 210 History of Liaison with Op-20-G in Washington. TNA HW 8/49. 1946, 46. 211 From Chalkley to Director, GCCS. July 1945, 1. 212 From Bodsworth to Director, GCCS. August 1945, 2.
196
underlying issue from contention, rather than retaliating or escalating the dispute.
From this point through the end of the war, all indications suggest that it was
‘business as usual’ between Op-20-G and the GCCS Special Liaison Unit.
* * * * * * * * * * *
In both of the above disputes, the British SLU personnel and their US counterparts placed a greater emphasis on ‘keeping the game going’ than in seeking to maximize their gains and retaliating when their unilateral aims could not be realized. This appears to have been true for even the most contentious disputes, such as Turing’s controversial visit to Bell Laboratories. This again provides a marked contrast with the Anglo-Soviet case study, in which 30 Mission argued for punishing their Soviet counterparts and “making them squeal” when the latter were perceived as having failed to deliver on their intelligence commitments.
Case prediction #3: The United States will meet Britain’s minimum standards of expected behavior as an intelligence partner. Comparatively speaking,
Britain expected far more of the United States as a military ally and intelligence partner than it did of the Soviet Union. These expectations—specifically concerning the types of information that would be shared, as well as the mechanisms by which it would be exchanged—had been spelled out in advance through formally negotiated agreements such as Travis-Wenger and BRUSA. These differences notwithstanding, the BLOs in Washington and Moscow shared common professional backgrounds and therefore had very similar ‘bare minimum’ expectations of what would constitute acceptable behavior from their hosts: direct access to their military service section counterparts, prompt responses to informational requests concerning the common
197
enemy, and sufficient latitude to carry out their official liaison duties from within the host country. I will now examine each of these expectations to determine whether they are borne out in the case.
3 (a) British personnel will be afforded direct, working-level access to their
American counterparts. As noted in the previous chapter, there was a prevailing view within the larger British military establishment that foreign intelligence liaison worked most effectively when officers were permitted to engage in direct dialogue with their functional counterparts. Evidence of this view is seen both in the private observations of 30 Mission personnel stationed in Moscow, as well as in the SLU’s monthly off-the-record notes from Washington.213
One key difference should be noted, however, when comparing the intelligence functions of 30 Mission and the SLU. Whereas 30 Mission personnel were responsible for exchanging finished intelligence with the Soviet General Staff
(e.g., assessments of German military capabilities and Order of Battle on the Eastern
Front), the SLU was responsible for sharing, exploiting and analyzing raw SIGINT traffic.214 Without the ability to engage directly and continuously with US SIGINT personnel at the working level, the SLU recognized that it would be unable to provide
GCCS headquarters with a complete picture of the Americans’ evolving cryptologic posture against Japan—let alone augment, complement and guide these US efforts with meaningful contributions of its own.
213 Major Geoffery G. (G.G.) Stevens wrote in November of 1942 that, based on his experience as a senior BLO assigned to Arlington Hall, direct discussion with working-level personnel was preferable to “the more impersonal type of exchange” because it “puts the liaison on a more personal basis between the two book-breaking sections and …lead[s] to a better understanding and liaison” (From Maj. Stevens to Cdr. Denniston. Untitled Letter. TNA HW14/57. November 6, 1942, 1- 2). 214 History of Liaison with Op-20-G in Washington. TNA HW 8/49. 1946, 18-19.
198
Britain’s expectations came into greater focus during the winter months of
1943 to 1944, when GCCS began implementing in earnest the geographic division of effort that had been negotiated in the May 1943 BRUSA Agreement. Hugh Foss, representing the Naval Section of GCCS, arrived in January of 1944 with the explicit mandate of “conduct[ing] Japanese naval technical (cryptographic) liaison on the working level” at Op-20-G.215 In a similar fashion, Wing Commander F. H. Culpin wrote that his primary objective as an integrated member of G-2 Special Branch was to liaise with multiple US military organizations “in all intelligence matters concerning the Japanese Air Force.” In order to carry out his assigned duties, Culpin noted that he would require direct “access to both the U.S. Army G-2 Air and A-2
Sections and the U.S. Navy ONI.”216
Evidence suggests that each of the SLU detachments was generally satisfied with the level of working-level access afforded by its US host agency. At the G-2
Special Branch, for example, Lieutenant Colonel Godley reported favorably on the manner in which he had been integrated into the operations of the Far Eastern Branch:
…[S]ince my arrival everyone with whom I have been in contact has been most helpful, kind, and completely willing to allow me to participate fully in the work of the Branch. In this way although I have not yet been at work in the Branch for 14 days I already feel an integral part of it and my work, suggestions and criticisms are accepted as from within.217
Godley’s assistant Captain L. J. Burrows echoed these sentiments, asserting that his current working arrangements ensured that “no Ultra intelligence of value from U.S.
215 Ibid., 25. 216 From Wing Cdr. Culpin to Director, GCCS. “Japanese Air Section Notes for February.” TNA HW 57/2. March 1944, 1. 217 From Godley to Director, GCCS. “Japanese Military Section Notes for February.” TNA HW 57/2. March 1944, 10.
199
military sources is failing to reach G.C.C.S.”218 This level of access enabled both men to observe and provide highly detailed reports on the composition, structure, roles, personnel and performance of the American G-2 Special Branch. It also gave them an opportunity to assess—through direct observation—the strengths and weaknesses of specific US SIGINT personnel with whom the British SLU was working during the war.
The Americans extended similar courtesies to the SLU detachments at Op-20-
G and SSA. The aforementioned Mr. Foss, for example, was allowed to make frequent visits to “all the relevant parties in the [Naval Communications] Annex, getting to know the people and becoming acquainted with their methods.”219 Major J.
R. Cheadle, who was part of the SLU detachment to Arlington Hall, noted that he felt
“completely at liberty to go to talk to Heads of Sections without going through the
Head of the Division or Branch Chief.”220
There were nevertheless instances in which SLU personnel encountered obstacles to meeting with their US counterparts. This appears to have occurred primarily on the naval side, where Op-20-G adhered to a more protocol-conscious mindset than the other US SIGINT agencies. One example of this occurred in
February of 1944, when Wing Commander Culpin noted that Op-20-G was only allowing him to visit when he had "some specific point to raise" and had made an appointment in advance.221 Another example is seen in the unsuccessful efforts of
218 From Capt. Burrows to Director, GCCS. “Jap. Military Section Notes for Nov. 1944 – Notes from Capt. L.J. Burrows.” TNA HW 57/2. December 1944, 12. 219 “History of Liaison with Op-20-G in Washington.” TNA HW 8/49. 1946, 12. 220 From Cheadle to Director, GCCS. July 1945, 1. 221 From Culpin to Director, GCCS. March 1944, 3-4; From Wing Cdr. Culpin to Director,
200
Lieutenant Chalkley and other British intelligence officers to gain access to the
Combat Intelligence (GI) branch of the Pacific Strategic Intelligence Section (PSIS) from January through May of 1945.222
The SLU ultimately decided against pursuing either of these matters any further. In the case of Culpin, the SLU reached an agreement with Op-20-G provide him a desk in their new facility, along with the privilege of “full and daily access to all material” that he might need.223 In the case of Chalkley, the SLU chose to cut its losses and transfer him to the Shipping Section of MIS (previously known as G-2
Special Branch), where he went on to serve as the Naval Section’s official representative.
Consistent with the dyadic democratic peace proposition, the SLU appears to have been largely content with the level of access provided to its BLOs in
Washington. In cases where the level of access was not fully satisfactory, the SLU was able to address the situation by negotiating for improved outcomes (as in the case of Culpin) or pursuing alternative arrangements (as in the case of Chalkley).
3 (b) Informational requests on matters concerning the common enemy will be answered in a prompt and satisfactory fashion. As described above, a series of formal agreements proclaimed that Britain and the United States had been put on a firm path toward full exchange and collaborative exploitation of Axis-related
SIGINT. In practice, this vision proved more difficult to realize than initially expected. The growing volume of daily SIGINT traffic, cryptographic recoveries and
GCCS. “Japanese Air Section Notes for March.” TNA HW 57/2. April 1944, p. 2-3. 222 “History of Liaison with Op-20-G in Washington.” TNA HW 8/49. 1946, 40-44. 223 From Culpin to Director, GCCS. March 1944, 3-4; From Culpin to Director, GCCS. April 1944, 2-3.
201
intelligence reports put significant stress on an already overloaded secure communications circuit (i.e., the Washington-GCCS radio teletype link), as well as on the US personnel who were responsible for manually compiling all of this material
(Erskine 2006, 7).
These constraints were not always fully appreciated at GCCS headquarters in
Bletchley Park. For example, there seems to have been a perception that the US
SIGINT agencies, with their ample technical resources and staffing, should have no difficulty providing prompt and satisfactory responses to routine British informational requests. In response to an accusation from Bletchley Park that Op-20-G was failing to deliver on its commitments (and thus operating outside the boundaries of expected behavior), GCCS Naval Section representative Hugh Foss countered that the US
Naval SIGINT organization was doing as much as could realistically be expected of it under the circumstances:
I am very sorry you feel that way. I think you are wrong. … In B.P. you may think of Op-20-G as so hopelessly over-staffed that they could easily cope with our very moderate requests but let me assure you that it is not so. …[T]hey rush this way and that so rapidly for their various processes that considerable ingenuity is required to catch them standing long enough to be photographed. I have been spending a week chasing JN 11224 traffic so I can speak from personal experience. … I feel I must tell you about this to emphasize that such requests are not granted with a smile and wave of the hand nor refused from motives of avarice, jealousy or sheer cussedness.225
Foss was optimistic that Op-20-G would eventually become more responsive to British informational requests. His successor, William Bodsworth, seemed to validate that prediction when he noted that Op-20-G was starting to come forward
224 JN-11, also known as the Maru Code, was a Japanese cipher utilized for the transmission of information concerning the movements of Japanese merchant vessels (Rohwer 1994, 547). 225 From Mr. H.R. Foss to Director, GCCS. “Op-20-G Notes for April 1944.” TNA HW 57/2. May 1944, 3.
202
with its own ideas for improving the overall handling of informational requests from
GCCS.226
An additional complicating factor was the lack of common, well-defined procedures for organizing and indexing the vast array of SIGINT-related material at
Op-20-G. The significance of this problem is seen in Bodsworth’s observation that
“the main delay here in getting replies is not means of communications but in finding and checking the information if is not available at once.”227 Since most informational requests were retrospective in nature (e.g., addressing collection statistics over some period in the past), it was important that the necessary data be properly indexed and filed ahead of time for optimal retrieval. When these conditions were not satisfied up front, answers had to be gathered and synthesized from scratch—a process that invariably resulted in further delays.228
In spite of these complications, there was a perception within the British SLU that the US SIGINT host agencies were acting in good faith and doing a creditable job of addressing British informational requests.229 When delays occurred, the SLU detachments tended to rationalize them as the natural consequences of a highly taxed
US SIGINT system or, alternatively, recast them as marginal improvements over the
226 From Bodsworth to Director, GCCS. “Op-20-G Notes for December 1944.” TNA HW 57/5. January 1945, 1. 227 From Bodsworth to Director, GCCS. “Op-20-G Notes for April 1945.” TNA HW 57/5. May 1945, 2. 228 From Lewis to Director, GCCS. July 1944, 1; From Chalkley to Director, GCCS. March 1945, 1. 229 Bodsworth, for example, praised Op-20-G for being “quick to act on any requests that are made through [him]” and expressed hope that GCCS was “being as well served” by US SIGINT liaison officers at Bletchley Park (From Bodsworth to Director, GCCS. “OP-20-G Notes for November 1944.” TNA HW 57/2. December 1944, 1).
203
previous year’s situation. To the extent that the SLU representatives expressed disappointment with anyone on the US side, it was with the top levels of military leadership, where there was less enthusiasm for Anglo-American intelligence cooperation. This viewpoint is reflected in the following post-war assessment provided by several members of the SLU detachment at Op-20-G:
A small group of Annapolis men with pre-war experience in Sigint were in complete control of all policy, and with one exception, occupied all the important positions. … As the organization expanded to take in a very large number of young, able and enthusiastic reserves, who were quick to learn and impatient of conservative methods, this oligarchy felt its position challenged, and was forced to defend itself by appealing to all the apparatus of naval hierarchy, rank, security regulations, insistence on the ‘correct channels,’ emphasis on experience, and conventional, time-approved methods. The rigid, tightly compartmentalized, strictly hierarchical organization which had to be maintained by the U.S.N. group in order to defend its position presented a very formidable problem to the outsider, intent on establishing close relations with those who really did the work (HW 8/49, “History of Liaison with Op-20-G,” 26 March 1946, p. 50).
It is difficult to determine conclusively whether US SIGINT agencies provided prompt and satisfactory responses to British informational requests. While it is true that Britain did not receive everything they requested from the United States, it is also true that British expectations were gradually tempered as they came to appreciate the operational constraints facing Op-20-G, SSA and the G-2 Special
Branch. The SLU detachments thus came to take a more flexible view of what constituted acceptable behavior from their US counterparts in this area. That flexibility was fueled by two factors: an evolving understanding of what was technically and politically possible under very challenging circumstances, as well as a desire to keep the game of Anglo-American intelligence cooperation going.
204
3(c) The US will afford British personnel sufficient latitude to carry out their official liaison duties at their assigned organizations. In contrast to the 30
Military Mission in Moscow, the SLU detachments operated in a much more open and tolerant environment. A thorough examination of the detachments’ monthly off- the-record notes suggests that the BLOs did not face many of the obstacles that 30
Mission experienced, such as being placed under constant surveillance and facing the prospect of domestic prosecution. At the same time, however, these BLOs were mindful of their status as guests in the national capital of a military ally. They recognized that they did not have carte blanche to operate within their assigned US host organizations, and that certain protocols had to be respected within each of the military service branches. The BLOs also understood that they would probably never obtain full and complete access to every item and resource that they requested of Op-
20-G, Arlington Hall and G-2 Special Branch.
What, then, constituted ‘sufficient latitude’ for the SLU detachments to perform their official liaison and cryptanalytic duties? A comprehensive examination of the monthly off-the-record notes reveals two overarching themes:
• The ability to serve as fully integrated ‘participant-observers’ within their assigned organizations. The BLOs sought to become involved directly in US cryptologic activities, as well as observe and report on these and related developments (e.g., leadership changes and intra-organizational politics) to GCCS Headquarters and Bletchley Park.
• Reliable access to facilities and resources as required in the course of performing regular duties. As guests of the US Government, the SLU detachments were largely dependent upon their hosts for facilities (e.g., office space), communications infrastructure (e.g., secure radio circuit) and supporting staff (e.g., typists and administrative support) required to do their jobs.
205
A strong case can be made that the US SIGINT agencies went a long way toward satisfying both expectations. To begin with, a number of SLU personnel appear to have been granted ‘participant-observer’ status within their US host organizations. This is seen in glowing descriptions such as the one provided by
Captain Burrows of his role within the Pacific Order of Battle Section of G-2 Special
Branch:
Every facility has been provided by Col. Perry and the Pacific O.B. Officers, who co-operated to the full in furthering the selection and passage of intelligence of value to G.C.C.S. and MI 2. I have been allotted a desk in the Current Intelligence Section, with the officers from the former Immediate Reports. I am thus immediately informed of any important development in the O.B. field, and am able to compose cables to G.C.C.S. in the closest contact with the U.S. officers similarly serving their own overseas commands. On items of sufficient importance advance information can thus be sent, details following in a routine cable later. Also easily available are the records and opinions of all other sections of Pacific O.B. Contact is particularly close with the Pacific O.B. T.A. officers, with whom I have detailed discussion on the daily S.S.A. report before composing the daily signal of T.A. intelligence indications to G.C.C.S..230
In addition to granting ‘participant-observer’ status to fully integrated liaison officers such as Captain Burrows, the US host organizations often extended similar privileges to temporary duty (TDY) personnel from Britain. Two visiting members of the Naval Section of Bletchley Park, for example, were temporarily integrated into various sub-sections of Op-20-G to support ongoing US efforts against Japanese
Naval ciphers and ‘Y’ traffic, respectively.231 Likewise, it was not uncommon for US
SIGINT agencies to host SLU personnel from other Washington-area detachments for regular visits. In one notable example, Major Cheadle (the senior BLO stationed at
230 From Burrows to Director, GCCS. December 1944, 12. 231 From Bodsworth to Director, GCCS. “Op-20-G Notes for September 1944.” HW 57/2. October 1944 1.
206
Arlington Hall) visited Op-20-G three times per week in order to remain “abreast of the cryptographic problems” and “maintain useful contacts” with the leadership there.232
The SLU detachments were generally less satisfied with the level of administrative support provided by their US host agencies. This view seems to have been based upon a perception that the resource-rich Americans could afford to do more for the British in this area, but for unspecified reasons chose not to. While not all of the SLU detachments shared this view,233 at least one of them did: the unit assigned to Op-20-G. In addition to expressing its frustration with the Navy’s limited use of the BRUSA circuit for transmitting secure communications (thus resulting in a diminished flow of raw traffic and operational cables to GCCS), the unit also accused
Op-20-G of “providing facilities to the British [only] when no inconvenience…was involved” and seldom “provid[ing] the B.L.O. with more than part-time secretarial assistance, even when asked to do so.” According to the official GCCS history of
SIGINT intelligence liaison with Op-20-G, it was not until the senior GCCS representative in Washington delivered “a strong démarche” that “the status and quarters of the British representatives were improved.”234
Following the démarche, the senior BLO at Op-20-G noticed a qualitative improvement in the level of support provided by his US hosts:
232 From Chalkley to Director, GCCS. July 1945, 1. 233 The Japanese Army section at G-2 Special Branch, for example, reported with some enthusiasm that it had been relocated to “a very agreeable corner…[with] an enormous map of Japan” as well as their own stenographer and close proximity to the Pentagon’s Intelligence Library (Sq/Ldr. B. B. King. “Japanese Army/Air Notes for November 1944.” HW 57/2. December 1944, 1). 234 “History of Liaison with Op-20-G in Washington.” 1946, 20.
207
News from here is good. Last month I reported that Poeder and Braun had begun to note the needs of the British Liaison and yours … and to make a serious effort to meet them. … Roeder has further noticed that a full time secretary should be provided for the British Liaison. Since the November move my part time typist has worked in another room. … If Roeder can persuade Ford to let me have this typist as full time secretary, it will lighten work considerably and help us to give you better service.235
While the SLU detachments did not get everything they sought from their
American host organizations, they appear to have been sufficiently well-served to carry out their official duties. There are likewise no indications that any of the US host agencies operated significantly outside the norms of bounded uncertainty. When
US support was found wanting in certain areas, the SLU detachments consistently worked within existing channels to compel improvements. When that approach did not work, they selectively invoked predictable and time-honored measures (e.g., the démarche) to prompt desired changes in behavior.
Testing the relational contracting argument
Measuring the dependent variable
Bindingness. In contrast to the previous chapter’s case study, the Anglo-
American WCID exhibited a very high degree of bindingness. During World War II
Britain and the United States concluded a number of agreements addressing various aspects of intelligence cooperation. As illustrated in Table 12, the clarity and formality of commitment of these agreements increased over time—beginning with general verbal agreements between senior military leadership and culminating with explicitly worded accords governing the mechanics of cryptanalytic exchange.
235 From Bodsworth to Director, GCCS. “Op-20-G Notes for January 1945.” HW 57/5, 1.
208
Table 12. Bindingness Matrix for the Anglo-American WCID, 1940-1945
Table 12 does not provide an exhaustive listing of every Anglo-American intelligence-related commitment during World War II, but it does it contain representative examples of the different types of intelligence commitments reached between the British and American Governments during this period. As in the previous chapter, each example is mapped to its appropriate position in the bindingness matrix presented in Chapter Three. Based on these examples, the Anglo-
American WCID appears to have been highly binding.
209
Measuring the independent variables
I will next measure the observed values of the independent variables for the relational contracting perspective that are expected to shape the bindingness of the
Anglo-American WCID.
Level of external threat. As discussed in the previous chapter, the presence of common background conditions in both cases—including a similar time period and common enemies—enables us to treat the level of external threat as an environmental constant. For purposes of this study, I consider it a controlled variable whose value
(extreme) does not affect relative variance over the dependent variable (bindingness) in the two cases.
Transaction costs. Consistent with the relational contacting perspective, the only remaining factor that could plausibly explain variance over the dependent variable is transaction costs. Earlier I operationalized transaction costs as a function of two proxies: asset specificity and heterogeneity of the participants. I will now consider each of these in turn.
Asset specificity. In contrast to the closed and highly restricted operating environment that 30 Mission personnel faced in Moscow, Washington, D.C. offered precisely the opposite for the SLU detachments: a relatively open atmosphere in which they were able to seek out and interact with their US counterparts at Op-20-G, the G-2 Special Branch and Arlington Hall. These relationships emerged as a natural outgrowth of the Travis-Wenger and BRUSA Agreements, which paved the road for working-level SIGINT liaison in both countries. From Britain’s perspective, the intrinsic value of these relationships hinged directly upon the active cooperation and consent of its American hosts. While the US military SIGINT agencies had agreed in
210
principle to ‘share everything’ with their British counterparts, there was always a risk that they could renege on their commitments—for example, by refusing to grant
GCCS access to certain information or US personnel. Thus, similar to the Anglo-
Soviet military liaison channels described in the previous chapter, the SLU detachments at Op-20-G, G-2 Special Branch and Arlington Hall represent examples of highly co-specific social capital assets.
A defining characteristic of a co-specific asset is that it cannot be easily removed from its current application and reallocated to some alternate use. Co- specific assets tend to exist solely within the unique context in which they were created. Once this foundational context begins to collapse, the risks of asset forfeiture increase. The Washington-based GCCS detachments understood this principle intuitively, as revealed in the following observation made by Captain
Burrows of the G-2 Special Branch detachment in his first monthly letter to the
Director of GCCS:
[W]hile the goodwill so generally created by Lt Col. Godley within the branch greatly smoothes the path of co-operation, the best results are unlikely to accrue from such a frequent change of [GCCS] personnel. Also in order to avoid dislocation in the smooth passage of intelligence to GCCS, an overlap of one month when a liaison officer is relieved is desirable.236
Burrows understood further that any reservoirs of goodwill that had been cultivated with the Americans by his predecessor (Lt. Col. Godley) would not last indefinitely.
If the SLU detachment at G-2 Special Branch wished to maintain the productive working-level relationships that it had cultivated during its first year in Washington, it would need to re-think its approach to managing personnel turnover—specifically by
236 From Burrows to Director, GCCS. “Notes for Monthly Letter, Military Section.” TNA HW 57/2. February 1945, 5.
211
giving the outgoing BLOs more time to introduce their replacements to the relevant
US SIGINT personnel and sections of Special Branch.
Is the asset What did the British capable of Relative Practical Comparative Asset Asset gain from its being ease of uses beyond value of category examples investment of this reallocated asset’s current alternate asset? to other withdrawal application? use(s) activities? Liaison Access to US SIGINT presence at personnel, raw traffic, Op-20G decrypts, intelligence Liaison reports, and presence at cryptanalytic methods Social G-2 Special No N/A N/A N/A capital Branch Potential to influence and guide the structure, Liaison composition and presence at management of US Arlington SIGNT organizations Hall
Reciprocal access to US Solutions of solutions of other major certain Intellec- Japanese cipher major tual systems, along with No N/A N/A N/A Japanese capital solutions to German cipher ciphers from Bletchley systems Park
Ability to engage with US counterparts at OP- GCCS Labor 20-G, G-2 Special Yes Easy personnel Yes Less Branch, and Arlington Hall
Table 13. Asset Inventory for the Anglo-American WCID: British Investments
In addition to social capital, Britain also invested substantial amounts of intellectual and labor capital in its wartime SIGINT relationship with the United
States. Table 13 depicts these investments in the context of a larger asset inventory for the Anglo-American WCID. There are several observations worth noting in this table:
• First, similar to the experience of 30 Mission, none of Britain’s social or intellectual capital assets could have been conceivably withdrawn and re- applied to alternate uses. This is illustrated in a simple counterfactual: assuming that the Anglo-American WCID had been nullified during the war, all of the social and intellectual capital assets which Britain had invested in the WCID up to that point (e.g., liaison relationships and Japanese cipher solutions shared with US SIGINT agencies) would have become sunken costs.
212
• Second, in contrast to 30 Mission, the SLU detachments did not share any weapons, equipment or matériel with their US SIGINT host agencies. There is, however, some evidence of the United States engaging in limited transfers of physical capital to GCCS throughout the war.237
• Third and finally, Britain arguably could have withdrawn its labor assets more easily from the Anglo-American WCID than from the Anglo-Soviet WCID. From Britain’s perspective, there were fewer viable employment alternatives for the Russia specialists stationed at 30 Mission than for the more general purpose SLU personnel assigned to Washington (the latter of whom could have been easily withdrawn and assigned to work on any variety of SIGINT problems).
Heterogeneity of states involved. The Washington-based GCCS representatives and their US counterparts had much more in common with one another than did 30 Mission and the Soviet General Staff. Most notably, they shared a common official language. Since all of the Anglo-American bilateral intelligence interactions were conducted in English, the two sides were much less vulnerable to the technical translation difficulties that had plagued 30 Mission in its efforts to exchange highly complex Chemical Warfare information with the Soviet General
Staff. To be sure, there arose instances in which SLU personnel and their US counterparts were (to invoke a phrase commonly attributed to George Bernard Shaw)
“separated by the same language,” as when the two sides utilized different terminology to describe the same technical concepts. As these semantic inconsistencies became more apparent over time, GCCS and the US SIGINT agencies worked together to resolve them through the creation of new institutions such as the
237 Among the more well-known of these transfers occurred in February of 1941, when a S.I.S. (Arlington Hall) delegation led by Army Captain Abraham Sinkov and Lieutenant Leo Rosen traveled to Bletchley Park and supplied (among other things) a working copy of the Japanese “Purple” cipher machine to their British hosts (Benson 1997, 19). The United States subsequently shipped as many as nine ‘Bombes’ (electro-mechanical machines used to decipher German-enciphered signals) to the United Kingdom for use by the aforementioned BEECHNUT units, which worked alongside GCCS in intercepting and decrypting German communications.
213
combined Terminology Committee led by William Friedman of Arlington Hall, as well as through the publication of technical manuals in which analogous American and British terms were printed adjacent to one another.238
The two parties also shared the distinction of serving fully democratic governments. While this was seldom acknowledged explicitly by the participants themselves, there is implicit evidence that their common form of government may have helped to reduce the transaction costs of bilateral intelligence cooperation.
Whereas the members of 30 Mission operated in a relatively mysterious and often frustrating political environment (as seen in its tense dealings with arcane institutions such as the Otdel and the Moscow civil government), the SLU detachments worked within an American military intelligence bureaucracy that was structurally and functionally similar to their own. As a result, they tended to have a much sounder grasp and appreciation of what was happening around them. This is revealed in the private writings of individuals such as Major J. R. Cheadle, who attended the inaugural Combined Signal Intelligence Meetings (CSIMs) at Arlington Hall towards the end of the war:
I feel that these meetings are very useful in showing general policy trends and enabling S.S.A. to ‘get off their chests’ any points on which they are unhappy and thus should serve to prevent minor points from becoming major issues. Also, as the Americans sometimes discuss points among themselves at these meetings, the British representatives can sit back and watch how the wind is blowing.239
As a representative of a fully liberal democratic government, Cheadle was already familiar with the role of such public forums in enabling participants to vent their
238 From Laslett to Director, GCCS. “Op-20-G Notes for May 1945.” HW 57/5. June 1945, 4- 5; From Cheadle to Director, GCCS. “Arlington Hall Notes for March 1945.” HW 57/5. April 1945, 2. 239 From Cheadle to Director, GCCS. July 1945, 2.
214
individual concerns and forge intra-organizational consensus. Thanks in part to this familiarity, Cheadle and his GCCS colleagues had less difficulty operating within and making sense of their operating environment, compared to their 30 Mission counterparts.
On balance, the members of the Anglo-American WCID faced lower net transaction costs than those of the Anglo-Soviet WCID. This is most evident from the standpoint of participant heterogeneity. Whereas the members of 30 Mission and
SGS struggled to overcome pronounced differences in language and political background, the SLU detachments encountered no such difficulties in their dealings with the US SIGINT agencies. With regard to asset specificity, however, the distinction between the two cases is less clear-cut. Both WCIDs were characterized by highly asset-specific investments of social and intellectual capital, neither of which could have been plausibly recouped in the event of defection by one or both members. Nevertheless, in terms of relative labor capital investments, the general- purpose SLU personnel were marginally less asset-specific (and thus more easily withdrawn and applied to other SIGINT activities) than the more specialized Russian subject matter experts of 30 Mission.
* * * * * * * * * * *
The relational contracting hypothesis posits a direct relationship between the level of threat and/or transaction costs incurred by a WCID and the degree of bindingness observed between its members. Given the relatively higher levels of transaction costs faced by 30 Mission and the Soviet General Staff, we should expect to see higher levels of bindingness in the Anglo-Soviet WCID than in the Anglo-
215
American WCID. That expectation is not supported empirically. Whereas the
Anglo-American WCID evolved into a highly binding relationship governed by explicitly worded agreements, the Anglo-Soviet WCID never evolved much beyond ad hoc intelligence cooperation. Of the few understandings that were reached between the British and Soviet Governments, none exhibited anything approximating the formality or clarity of commitment of the Travis-Wenger and BRUSA agreements.
The observed values of the dependent and independent variables are therefore inconsistent with the relational contracting hypothesis. This casts some doubt on the ability of this perspective to explain variance in bilateral intelligence outcomes. In an effort to extend this plausibility probe further, I will next consider some of the observable implications of the relational contracting perspective and test for evidence of their presence within the case.
Testing for observable implications
As noted in Chapter Three, Weber’s anarchic perspective does not account for the possibility of hierarchy as a bilateral cooperation outcome. This becomes problematic when we attempt to extend her relational contracting argument to the study of intelligence cooperation, where hierarchical outcomes are more prevalent. I suggested earlier that we could help to address this problem by incorporating aspects of Walsh’s willful hierarchy argument to generate observable implications of relational contracting within both cases.
Walsh’s argument is a suitable choice because it makes highly specific and contingent predictions about how binding intelligence relationships will materialize, given the presence or absence of three key background conditions: high expected
216
gains of intelligence cooperation, fears of defection and the presence of a power imbalance between the parties. When all three conditions are satisfied, the parties are expected to forge a binding relationship in which the dominant party assumes a degree of control over the subordinate party’s intelligence activities. Alternatively, when the parties’ fears of defection are negligible, they are expected to share intelligence through anarchic institutions that preserve their decision-making authority (e.g., bilateral governance structures).
Before we can generate observable implications of relational contracting within the Anglo-American case, we must first determine whether each of the three aforementioned background conditions is present. The answers to these questions will help to determine what outcomes we should expect to see in the case, assuming that the argument is true.
(1) Did both parties believe that intelligence cooperation against the common enemy would be fruitful? While its primary goal was to bring the US military into the war on the Allied side, Britain also saw the United States as providing potentially complementary intelligence capabilities (e.g., collection accesses, high-speed decryption capabilities and sheer manpower) against Japan, which had become a proverbial blind spot for GCCS during the early years of the war.
The United States, for its part, saw value in pursuing greater intelligence cooperation with Britain against Germany and Japan. This is reflected in the early decisions of high-ranking US military officials such as General George Marshall, who in September of 1940 approved the G-2 Special Branch to serve as the War
Department’s official liaison for coordinating intelligence exchange with members of
217
the British Technical Mission and authorized representatives of the British Armed
Forces.240 Arlington Hall likewise saw significant benefits in closer cooperation with the British, such as sending junior Army personnel to work under senior cryptanalysts at Bletchley Park and drawing upon their extensive background knowledge on enemy cipher systems dating back to the Spanish Civil War.241
(2) Did at least one of the parties have reason to suspect the other of defecting? Comparatively speaking, 30 Mission and the Soviet General Staff were far more concerned about the possibility of defection than the SLU detachments and their US SIGINT counterpart agencies. This is not to suggest, however, that the members of the Anglo-American WCID discounted the possibility of defection. On the contrary, each side was clearly concerned about the possibility of defection and took measures to guard against it.
This was certainly evident within the US SIGINT establishment. Each
American host agency kept close tabs on the comings, goings and activities of the
SLU personnel assigned to it. This included reading the senior BLOs’ on the record monthly liaison letters sent to Bletchley Park. By mutual agreement, the BLOs were supposed to file copies of these outbound reports with their US host agency— ostensibly to ensure that the latter’s leadership were kept apprised of by working- level developments that might later catch them off guard.242 These reports also
240 Brig. Gen. Sherman Miles. “Directive to G-2 Covering Interchange of Secret Technical Information with Representatives of British Government.” Item 4566. Box 1413. NACP RG 457. September 9, 1940, 1. 241 Maj. Solomon Kullback. “The British GC & CS.” Item 4565. Box 1413. NACP RG 457. August 1, 1942, 8. 242 W. Preston Corderman and H. M. O’Connor. “Summary of Notes Resulting from Discussions between British Representatives of Civil Branch, GC & CS and American Representatives of Signal Security Agency.” Item 2738. Box 940. RG 457. December 24, 1943. It should be noted that
218
provided the United States with continuous ‘on the record’ accounts of what British liaison officers were doing within their assigned host organizations—thus serving as a hedge, of sorts, against defection.243
A similar vigilance against defection was evident within the SLU detachments. As revealed in the monthly off-the-record notes, the BLOs were on a constant lookout for indications of their hosts failing to meet their obligations as an intelligence partner (e.g., as revealed in the dustup over the BRUSA secure communications circuit). The off-the-record notes also contain numerous candid assessments of US intelligence personnel, as well as accounts of US inter-service disputes and their potential implications for Anglo-American intelligence cooperation.
(3) Was there a clear dominant and subordinate entity in this relationship?
Despite possessing certain relative advantages over the United States in terms of its intelligence capabilities and subject matter expertise, Britain had fewer aggregate material resources at its disposal than the United States, on which it had grown dependent for economic and military support vis-à-vis Germany since the late 1930s
(Reynolds 1981, 117). This trend of rising US dominance continued well into the war, to the point that Brigadier General Vivian Dykes of the BJSM in Washington felt compelled to observe in 1942 that ‘there are few things which we are in a position
both parties were still able to route highly sensitive ‘off the record’ messages to their home organizations via separate channels. Evidence from the GCCS Naval Section detachment assigned to Op-20-G suggest that ‘these off the record’ accounted for a relatively small but significant proportion (approximately 23%) of total outgoing material sent to GCCS (From Bodsworth to Director, GCCS. “Reply to Editorial Comment on History of Liaison with Op-20-G (AZ/3601 of 3rd March refers).” HW 8/49. March 26, 1946, 9.). 243 Albert Howard Carter. “Liaison with the London Offices of GCCS.” Item 2741. Box 941. NACP RG 457. April 11, 1945, 1.
219
to give to the United States; but we are dependent on them for a great many things”
(Smith 1993, 116). The United States was thus by any conventional measurement the dominant party in the Anglo-American WCID.
* * * * * * * * * * *
Observable implications of the relational contracting perspective
(1) The dominant party will assume some measure of control over the subordinate party’s intelligence services.
(2) In exchange for its cooperation, the subordinate party will receive resources from the dominant party that exceed the value of shared intelligence
(3) The dominant party will pursue any of the following measures in its dealings with the subordinate party to reduce the likelihood of defection: (a) limit intelligence sharing to specific topics of common interest; (b) clearly specify the types and/or topics of intelligence that will (and will not) be shared; (c) monitor the subordinate state’s compliance with sharing arrangements; (d) punish any reneging by the subordinate party; and (e) restructure and/or encourage reform of the subordinate state’s intelligence services, such that they are more inclined to cooperate.
Table 14. Relational Contracting Argument: Anglo-American Case Predictions
Based on the answers to the above questions, we are once again presented with a set of observable implications favoring the creation of a binding hierarchical intelligence-sharing arrangement. The predictions, which are displayed in Table 14 above, are identical to those generated for the Anglo-Soviet WCID.
Case prediction #1: The dominant party will assume some measure of control over the subordinate party’s intelligence services. Contrary to the expectations of this argument, there is no evidence of the United States attempting to manage British intelligence activities or directly oversee their collection of enemy
220
communications. Each of the formal agreements described above treated the signatories as autonomous, co-equal partners engaged in a mutually beneficial geographic division of effort. These agreements did not contain any discussions of
‘price’ or references to ‘quid pro quos’ (e.g., conferring certain privileges to the dominant party in exchange for risk premiums awarded to the subordinate member).
Instead, they articulated the general security rules, principles, terminology and standards governing bilateral SIGINT cooperation, while at the same time entrusting the signatories to implement the provisions as they saw fit.
The Anglo-American WCID thus did not assume the form of a willful hierarchy. If anything, the case outcomes are far more consistent with anarchic sharing, given that the parties focused primarily on “developing efficient technical practices and standards for the exchange of information” rather than on curbing the risks of defection (Walsh 2010, 25). The willful hierarchy argument thus fails to offer a persuasive and empirically valid claim of the conditions under which a dominant intelligence organization is expected to assume control over a subordinate.
Case prediction #2: In exchange for its cooperation, the subordinate party will receive resources from the dominant party that exceed the value of shared intelligence. The United States supplied Britain with significant quantities of non- intelligence resources in the lead up to and during the Second World War, beginning with such well-known initiatives as the Bases-for-Destroyers Deal and Lend-Lease.
While collectively these transfers may have exceeded the value of intelligence shared within the Anglo-Soviet WCID, they were not conditioned explicitly upon submission to a subordinate role as an intelligence partner, as the willful hierarchy
221
argument predicts (Ibid., 23). The Anglo-American WCID was therefore not a complex liaison arrangement, defined earlier as one “involving the bartering of intelligence collection assets for some mix of political, intelligence, economic, military, or operational goods provided through intelligence channels” (Sims 2006,
197).
True or False for the Anglo- Claim Assessment American WCID? (a) Sharing is limited to Intelligence cooperation was limited to specific topics of common topics concerning the common enemy, with True interest and around which a specific emphasis on the Axis powers of both parties’ interests align. Germany and Japan. (b) The parties clearly At its core, the Anglo-American WCID specify the types and/or was an explicitly defined geographic True topics of intelligence that division of effort between the US and UK will be shared. military intelligence SIGINT services. Implicit monitoring was observed on both (c) The dominant state sides, as seen in the examples of Britain monitors the subordinate tracking America’s compliance with the Unclear state’s compliance with Naval BRUSA agreement and the United sharing arrangements. States counting the number of SLU ‘on the record’ reports filed with Arlington Hall. None of the formal bilateral agreements specify how defection was to be punished. (d) The dominant state In a handful of cases such as the CB 4377 introduces explicit False incident, the United States appears to have punishments for reneging. taken a harder line against the British for perceived acts of defection. Nevertheless, these sorts of retaliations were rare. In fact, it was the subordinate party that (e) The dominant state sought to influence changes in the restructures and/or structure, composition and organization of encourages reform of the the dominant party’s SIGINT subordinate state’s False organizations. This is seen most notably in intelligence services, such a series of recommendations made by Lt. that they are more inclined to Col. Lewis to the SSA leadership cooperate. concerning a substantial reorganization of their agency.244 Table 15. Expected Anglo-American Defection-Avoidance Strategies
244 From Lewis to Director, GCCS. “Report on Attachment to S.S.A. for July.” HW 57/2. August 1944, 8.
222
To the extent that bartering did occur between the SLU detachments and their
US host agencies, it generally assumed the form of a ‘simple’ liaison arrangement in which all bargaining occurred within intelligence channels. This is reflected further in the fact that the SLU detachments were created for the express purpose of facilitating bilateral intelligence exchange. Unlike 30 Mission, they did not have the authority to barter in both intelligence and military matériel.
Case prediction #3: The dominant party will attempt to reduce the likelihood of the subordinate party’s defection through any of a variety of measures. Similar to the previous chapter, we can test this prediction by considering
Walsh’s list of “five ways [that dominant states can] minimize the chances and costs of defection by other participants” and determining whether any evidence of these appears in the case (Walsh 2007, 162). These claims appear in Table 15 above, along with an assessment of whether they were observed within the relationship.
Conclusion
Following an initial period of limited and largely anarchic information sharing against Germany during World War I, the British and US intelligence services ultimately chose to go their separate ways, with each citing concerns about the other’s long-term suitability as an intelligence partner. The two sides began to consider reactivating those ties in the late 1930s with the rise of Imperial Japan and Hitler’s
Germany. What subsequently emerged was the most ambitious and wide-ranging intelligence-sharing arrangement ever enacted: an inter-service military intelligence relationship governed by an explicitly codified geographic division of effort and
223
detailed security guidelines concerning the handling, sharing and dissemination of high-grade SIGINT.
The two exploratory hypotheses examined in this chapter addressed separate but related aspects of the dependent variable, depth of intelligence cooperation.
Consistent with the expectations of the dyadic democratic peace argument and its regime recognition dynamic, we observed numerous examples of both sides adhering to the norms of contingent consent and bounded uncertainty. In examining the observable implications of this theory from the vantage point of the SLU detachments in Washington, D.C., we saw consistent evidence of the parties ‘keeping the game going’ and eschewing defection—even amidst high-level disputes and mini-crises, such as the CB 4377 incident. The BLOs also learned quickly what they could (and could not) reasonably expect from their US host agencies. For the most part, those expectations were satisfied—and in those rare instances in which they were not, the two sides proved willing and capable of reaching a mutual accommodation.
The relational contracting argument posited a direct and positive relationship between the level of external threat and transaction costs, on the one hand, and the degree of bindingness in the relationship, on the other. The expected relationship was not borne out in this particular case. Compared to the Anglo-Soviet WCID, the
Anglo-American WCID exhibited slightly lower transaction costs but a much higher degree of bindingness, as seen in the creation of highly formalized and explicitly worded agreements such as Travis-Wenger and BRUSA. Most of the willful hierarchy predictions for this case did not materialize either. Contrary to expectations, the dominant power in the relationship (the United States) did not
224
attempt to assert direct control over the subordinate’s (Britain) intelligence activities.
In some cases the subordinate power acted more like a dominant power, as seen in the efforts of GCCS liaison officers to influence fundamental reorganizations of US military SIGINT organizations.
Of the two theories examined in this chapter, the dyadic democratic peace proposition once again appears to offer a more effective explanation of the case- specific outcomes. The relationship between the dependent and independent variables is largely consistent with our predictions, as are most of the observable implications. The relational contracting perspective, by contrast, fails to predict or explain most of the Anglo-American case outcomes.
225
Chapter 6: Conclusion
Argument Summary
The problem
Bilateral intelligence cooperation has received increased attention and public exposure in recent decades. Interest in this ‘missing dimension’ of international politics has been fueled by a growing recognition of the role of intelligence-sharing in detecting and preventing imminent terrorist attacks, supporting multilateral military operations and enabling resource-constrained governments to leverage their finite intelligence capabilities more efficiently. Until recently, much of the scholarship on bilateral intelligence cooperation has consisted of atheoretical case studies and detailed historical accounts. Cases are often depicted as idiosyncratic events, rather than as instances of a larger phenomenon. As a result, deeper conceptual and methodological questions—such as what constitutes bilateral intelligence cooperation and how it can be measured over time—have gone largely unexplored.
More can be done within the academic literature to address these questions and, in the process, inform public understanding of this topic. This presents an opportunity to political scientists and IR theorists. As I argued in Chapter Two,
International Relations offers a robust conceptual toolkit that, with subtle modifications, could yield new insights into intelligence-related research questions.
To date, this toolkit has been underutilized due to four longstanding obstacles: the conflicting value systems of the academy and the intelligence community, the academy’s historical suspicions of the intelligence community, inadequate primary
226
source data, and methodological commitments favoring large-n statistical studies over small-n qualitative case studies. I contend that these obstacles are beginning to recede, thanks in part to a more liberal US declassification regime, the gradual entry of Intelligence Studies into the academic mainstream, and the recent improvement and subsequent rehabilitation of qualitative research methods.
Research question
In this dissertation I have attempted to address the question of why certain pairs of states engage in greater degrees of intelligence cooperation than others. I selected this question for several reasons. First, it is representative of a larger class of intelligence-related questions that would benefit from the application of a theoretically informed case design. Second, it is a ‘policy-relevant’ question, as evidenced by recent debates and discussion within the United States concerning the relative merits (e.g., specialization) and drawbacks (e.g., the snowing effect) of cooperating with autocratic intelligence services. Any prospective answers or insights generated in response to this question could hold implications for how democratic governments select and manage their bilateral intelligence relationships in the future.
Two theories
For this study I have drawn upon two mainstream IR theoretical traditions: the dyadic democratic peace argument and relational contracting. In Chapter Three I considered how each perspective could be adapted to explain distinct but related aspects of my dependent variable—depth of intelligence cooperation—which I define
227
as a function of three attributes: frequency of intelligence exchange, granularity of information exchanged, and degree of structural commitment.
My adaptation of the dyadic democratic peace argument draws upon the writings of Kant (1795), Doyle (1985, 1986 and 1989), Schmitter (2002) and Dixon
(1994). It predicts that the frequency and granularity of information exchanged should vary in accordance with the regime composition of the dyad. Fully democratic dyads (i.e., those in which both members represent fully democratic regimes) are expected to engage in more frequent and granular exchange with one another, whereas mixed dyads (i.e., those in which the members consist of an autocracy and democracy) are expected to engage in less of both over time. This occurs as the result of a Janus-faced regime recognition dynamic in which, via the causal mechanisms of contingent consent and bounded uncertainty, liberal democracies benefit from a presumption of amity whilst autocracies face a presumption of enmity when dealing with liberal democracies. In sum, fully democratic dyads are expected to behave in a way that keeps the game of intelligence cooperation going to one another’s mutual benefit. Mixed dyads, by contrast, are expected to be plagued by mutual concerns about defection, thus resulting in less frequent and granular levels of intelligence exchange.
The relational contracting perspective is informed by the transaction cost economics literature and subsequent IR adaptations by Weber (1997, 2000). It highlights the role of several exogenous factors in shaping the bindingness of an intelligence relationship: transaction costs, the level of external threat and environmental uncertainty. Pairs of states facing high values over these variables are
228
expected to abandon ad hoc modes of intelligence cooperation in favor of more binding arrangements (e.g., formal intelligence alliances) that they believe are less vulnerable to defection. By contrast, states facing relatively lower values over these variables are expected to favor ad hoc modes of intelligence cooperation wherever possible.
In an effort to identify the observable implications of the relational contracting perspective within my selected cases, I leveraged insights from Walsh’s willful hierarchy argument (Walsh 2007, 2010). Walsh identifies three additional case conditions, in addition to those cited by Weber, that influence the type of binding arrangement that the parties will select: high expected gains from intelligence cooperation, fears of defection on either side and the presence of a power imbalance.
When all three conditions are satisfied, the parties are expected to form a willful hierarchy in which the dominant state assumes direct control of the subordinate’s intelligence services in exchange for side payments or rents. Alternatively, when the parties’ fears of defection are negligible, they are expected to share intelligence through anarchic mechanisms (e.g., bilateral governance structures or ad hoc sharing) that preserve their decision-making authority.
Methodology and case selection criteria
Having adapted these two theoretical perspectives to the study of bilateral intelligence cooperation, I next sought to clarify my research objectives and case selection criteria. Here I faced several problems. First, given the theoretically impoverished state of Intelligence Studies discussed in Chapter Two, I felt that it would be analytically premature to follow the conventional approach of devising and
229
testing rival hypotheses that make mutually exclusive claims about so-called ‘crucial cases’ (e.g., in a ‘most similar’ or ‘least similar’ research design). I concluded that it would be more sensible to draw upon a combination of theory-development and theory-testing research designs. This hybrid approach allowed me to determine whether my selected theories could be suitably extended to a new problem domain, while at the same time enabling me to construct and implement preliminary tests (i.e., plausibility probes) of these arguments against some initial cases.
A second problem concerned case selection. As noted earlier, a key limitation of any intelligence-focused study is the dearth of readily available primary source data. In order to compensate for these limitations and implement my envisioned hybrid research design, I sought to identify two historical case studies for which available data was known to exist. Additionally, in an effort to mitigate against some of the more well-known problems associated with qualitative small-n studies (e.g., confounding and omitted variable bias), I identified and attempted to satisfy five additional case selection criteria: strive to hold constant as many competing independent variables as possible, incorporate war as a stress condition, control for realism, select cases in which the primary units of analysis perform similar intelligence functions, and select cases exhibiting extreme variance over the dependent variable.
The Anglo-Soviet and Anglo-American wartime cooperative intelligence dyads (WCIDs) came close to satisfying all five case conditions. Each involved a
British military intelligence liaison unit operating in the capital city of an Allied country during World War II, with the primary goal of developing an intelligence
230
exchange relationship against a common enemy. For each case I was able to identify a robust body of primary source information—the war diaries of the British inter- service 30 Military Mission to Moscow, for the former, and the off-the-record monthly notes of the Washington, D.C.-based Signals Intelligence Liaison Unit
(SLU), for the latter. Through an in-depth examination of both data sets, I was able to measure and compare the posited relationships of the dependent and independent variables identified in Chapter Three, as well as test for the presence of observable implications of both theories.
Case findings and theoretical implications
In Chapters Four and Five I placed the Anglo-Soviet and Anglo-American
WCIDs in their respective historical contexts and tested my adaptations of the dyadic democratic peace and relational contracting arguments against them. In this section I highlight the key findings from these cases, along with some of their potential theoretical implications.
(1) The dyadic democratic peace hypothesis (H1) was largely corroborated in both cases, thus suggesting that it warrants further consideration as a possible explanation of bilateral intelligence cooperation.
In the case of the Anglo-Soviet WCID, the values of the independent variables
(extremely high and low polity values for Britain and the USSR, respectively) and dependent variables (diminishing levels of frequency and granularity) were consistent with my expectations of a highly heterogeneous (read: mixed) dyad. The observable implications of this argument in this case were similarly consistent with my expectations. I found numerous instances of the Soviet General Staff (SGS) refusing
231
to ‘keep the game going’ with its 30 Mission counterparts. In addition, the highly erratic and unpredictable behavior of the Soviet Otdel most certainly did not meet 30
Mission’s minimum standards of acceptable behavior for an intelligence liaison partner. As the war progressed, both of these factors contributed to an increasingly poisoned atmosphere of rapidly diminishing intelligence cooperation.
The only unexpected finding was the failure of 30 Mission to pursue retaliatory measures against the SGS and Otdel, as predicted by the theory. As I noted in Chapter Four, this discrepancy had nothing to do with the attitudes and preferences of 30 Mission, which was keen to invoke such measures throughout its tenure in Moscow. Rather, it reflects the realities of complex liaison within a bilateral intelligence relationship. The British Government’s refusal to endorse 30 Mission’s proposed retaliations can be explained as a result of overriding strategic calculations—namely, its desire to minimize and pre-empt any distractions that might detract from the Red Army’s strong fighting performance against Hitler’s forces on the Eastern Front.
In the case of the Anglo-American WCID, the expected values of the independent and dependent variables were also consistent with my expectations of a fully democratic dyad. Following the signing of the 1943 BRUSA Agreement, representatives from the primary British and American military SIGINT agencies moved quickly to establish integrated intelligence operations in Britain and the
United States. By late December, senior representatives from Britain’s General Code and Cypher School (GCCS) had arrived in Washington, D.C. to form the SLU and establish an embedded liaison presence within the Navy’s Op-20-G and the Army’s
232
G-2 Special Branch and Signal Security Agency (SSA) at Arlington Hall. From that point through the end of the war, a senior British liaison officer (BLO) and complementary cryptologic personnel were present at all three agencies, where they had largely uninterrupted access to their American counterparts and US SIGINT facilities. This constituted a marked contrast with the Anglo-Soviet WCID, in which the members of 30 Mission were only authorized to meet with their SGS counterparts in pre-arranged meetings to discuss pre-determined topics under the watchful eyes of
Soviet minders.
The observable implications of the dyadic democratic peace argument were also borne out in the Anglo-American WCID. To the extent that disputes arose in the course of day-to-day intelligence interactions, the SLU personnel and their US counterparts reached mutual accommodations that ‘kept the game going’ without jeopardizing any of the benefits of intelligence cooperation. Likewise, when certain intelligence agreements or outcomes favored the US SIGINT agencies, the SLU remained secure in its belief that it could continue to influence the dynamics of the overall relationship without fear of retaliation from their US hosts..
In sum, the modified version of the dyadic democratic peace argument was found to provide accurate predictions of the outcomes and mechanics of bilateral intelligence cooperation in two distinctly different cases occurring under the highly stressful background conditions of total war against a common enemy. The ability of the argument to perform effectively under these circumstances suggests that there is merit in extending the dyadic democratic peace perspective to explain other known cases of bilateral intelligence cooperation.
233
(2) Scholars can further test the strength, scope and applications of the dyadic democratic peace perspective by selecting new cases that control for the effects of potentially confounding variables (e.g., official language utilized by the parties), as well as those that vary in terms of their background conditions.
The cases selected for this study were admittedly imperfect, in the sense that both contained at least one potentially confounding variable (official language) that may have influenced the outcomes. It is possible, for example, that the shared
English language of the British and US Governments may have contributed disproportionately to the deeper levels of cooperation observed in the Anglo-
American WCID. Conversely, the lack of a shared official language between the
British and Soviet Governments may have contributed to the lower levels of cooperation seen in that relationship. This problem could be addressed by selecting fully democratic intelligence relationships in which the participants do not share a common official language (e.g., Germany and the United States since 2001) and mixed dyads that do (e.g., the United States-Pakistan dyad of more recent years).
Scholars may also wish to explore whether the dyadic democratic peace perspective can successfully explain variation in the depth of bilateral intelligence cooperation under less stressful conditions short of world war. For example, the regime composition of a dyad may prove less important as an explanatory variable when the stakes of cooperation are lower and the background conditions of the cases are allowed to vary. As new cases meeting these criteria are identified and examined, it will also be important to consider the possibility of complex interaction dynamics
(e.g., regime composition interacting with some other factor) affecting the observed
234
outcomes. The development and creation of the large-n USIIB statistical dataset could help to facilitate this sort of controlled testing, as one scholar has recently demonstrated (Tuzuner 2009).
(3) The relational contracting hypothesis (H2) was not well supported in either case, thus suggesting that the argument and its purported causal mechanism of hierarchy may be overstated.
In the case of the Anglo-Soviet WCID, the values of the independent variables
(extreme level of external threat, high uncertainty and high transaction costs) and dependent variable (low bindingness of intelligence cooperation) were inconsistent with my expectations of a highly heterogeneous dyad. The observable implications of the relational contracting argument also did not unfold as expected. Contrary to the posited case predictions of the relational contracting argument, the high transaction costs of intelligence cooperation facing Britain (e.g., 30 Mission’s difficulties operating within an autocratic environment and its need for translation services) along with its inability to recoup highly asset-specific investments in the relationship (e.g., the topic-specific meeting channels) did not lead to the development of a willful hierarchy. This failed to occur in spite of considerable side payments made by the British to the Soviet Union in the form of military materiel and logistical support.
In the case of the Anglo-American WCID, the values of the independent variables (extreme level of external threat, high uncertainty and relatively high transaction costs) and dependent variable (high bindingness) were consistent with the posited relationship in H2, so far as it goes. Nevertheless, a comparison of the Anglo-
235
American case outcomes with those of the Anglo-Soviet WCID suggests the possibility of omitted variable bias. Recall that two of the independent variables in the relational contracting argument—level of external threat and uncertainty—were treated as environmental constants because their values were more or less identical for the duration of both cases. Transaction costs thus served as the only remaining explanatory variable for the relational contracting argument. Given that transaction costs were lower for the Anglo-American WCID than for the Anglo-Soviet WCID, it does not make sense (per the logic of relational contracting) that the level of bindingness proved to be so much higher for the former than in the latter. This leaves open the possibility of another variable—one not accounted for explicitly in the relational contracting argument—doing the proverbial heavy lifting.
While several of the willful hierarchy case predictions for the Anglo-
American WCID were borne out in the case, the most important one was not. At no point in their wartime intelligence relationship did the United States (the dominant party) attempt to assume direct control of Britain’s (the subordinate state) intelligence activities. Moreover, in contrast to the Anglo-Soviet case, there does not appear to have been any pressure for such a take-over to occur from within the US military intelligence establishment—nor pressure to stop it from senior US leadership.
(4) Future scholarship will benefit from a broader consideration of the conditions under which hierarchical intelligence relationships (willful or not) emerge.
Walsh’s hierarchical extension of the relational contracting perspective is predicated on the notion of a less powerful state submitting willfully into an intelligence hierarchy with a more powerful state, typically in exchange for side
236
payments and other assurances against defection (Walsh 2007, 164). It is nevertheless conceivable that the decision to enter into a hierarchical intelligence relationship is not always a willful choice on the part of the weaker party. Some of the best-known cases of bilateral intelligence hierarchies involve dyads in which the weakest member was already subordinate to the dominant party (i.e., well before the onset of intelligence cooperation). Two of the examples cited by Walsh in his 2010 book—the United States and West Germany during the 1950s and the United States and South Vietnam during the 1960s and early 1970s—certainly seem to fit this mold
(Walsh 2010, 31-33 and 59-78).
The Anglo-Soviet case study does not fit this mold, however. Despite the presence of a power imbalance between the two countries, the weaker party (the
Soviet Union) did not submit to the dominant party (Great Britain) at the time they initiated intelligence cooperation. As scholars seek to extend and build upon the relational contracting perspective as an explanation of intelligence outcomes, they may wish to consider how often—and under what conditions—a less powerful state that is not currently subordinate to the more powerful state will enter willfully into a hierarchical intelligence relationship.
Conclusion
The theoretical arguments and cases considered in this study do not constitute a sufficient basis upon which to make broader generalizations about the role of regime composition or transaction costs in shaping the depth of cooperation in contemporary bilateral intelligence relationships. As of this writing, the state of theory development within the Intelligence Studies literature is just getting off the
237
ground—and our knowledge of the larger universe of cases remains too limited—to entertain such ambitions. I have attempted to make a modest contribution to this literature by joining an emerging group of scholars that is attempting to bring
Intelligence Studies—for decades a ‘missing dimension’ and ‘neglected colony’— back into the mainstream of International Relations. I have done so specifically by:
• Developing a novel extension of a well-established IR theoretical tradition (the dyadic democratic peace argument) and applying it directly to the study of bilateral intelligence cooperation.
• Exploring the plausibility of this argument, alongside another theoretical tradition (relational contracting) that has previously been applied to the study of intelligence, by ‘going deep’ into the details of two historical cases and determining whether the expected relationships and outcomes were present.
• Suggesting concrete ways in which both arguments could be fruitfully extended and improved upon in future academic work.
238
Appendix I: Analytic Framework for Bilateral Intelligence Cooperation
239
Appendix II: Twenty-Year Mutual Assistance Agreement Between the United Kingdom and the Union of Soviet Socialist Republics (1942)245
London, May 26, 1942
HIS MAJESTY THE KING OF GREAT BRITAIN, IRELAND AND BRITISH DOMINIONS BEYOND THE SEAS, EMPEROR OF INDIA, AND THE PRESIDIUM OF THE SUPREME COUNCIL OF THE UNION OF SOVIET SOCIALIST REPUBLICS:
Desiring to confirm the stipulations of the agreement between His Majesty's Government in the United Kingdom and the Government of the Union of Soviet Socialist Republics for joint action in the war against Germany signed at Moscow, July 12, 1941, and to replace them by formal treaty;
Desiring to contribute after the war to the maintenance of peace and to the prevention of further aggression by Germany or the States associated with her in acts of aggression in Europe;
Desiring, moreover, to give expression to their intention to collaborate closely with one another as well as with the other United Nations at the peace settlement and during the ensuing period of reconstruction on a basis of the principles enunciated in the declaration made Aug. 14, 1941, by the President of the United States of America and the Prime Minister of Great Britain, to which the Government of the Union of Soviet Socialist Republics has adhered;
Desiring finally to provide for mutual assistance in the event of attack upon either high contracting party by Germany or any of the States associated with her in acts of aggression in Europe;
Have decided to conclude a treaty for that purpose and have appointed as their plenipotentiaries:
His Majesty the King of Great Britain, Ireland and the British Dominions Beyond Seas, Emperor of India, for the United Kingdom of Great Britain and Northern Ireland:
The Right Hon. Anthony Eden, M. P., His Majesty's Principal Secretary of State for Foreign Affairs;
The Presidium of the Supreme Council of the Union of Soviet Socialist Republics:
245 The Avalon Project, 2008.
240
M. Vyacheslaff Mikhailovitch Molotoff, People's Commissar for Foreign Affairs,
Who, having communicated their full powers, found in good and due form, have agreed as follows:
PART ONE
ARTICLE I
In virtue of the alliance established between the United Kingdom and the Union of Soviet Socialist Republics, the high contracting parties mutually undertake to afford one another military and other assistance and support of all kinds in war against Germany and all those States which are associated with her in acts of aggression in Europe.
ARTICLE II
The high contracting parties undertake not to enter into any negotiations with the Hitlerite Government or any other government in Germany that does not clearly renounce all aggression intentions, and not to negotiate or conclude, except by mutual consent, any armistice or peace treaty with Germany or any other State associated with her in acts of aggression in Europe.
PART TWO
ARTICLE III
1. The high contracting parties declare their desire to unite with other like-minded States in adopting proposals for common action to preserve peace and resist aggression in the post-war period.
2. Pending adoption of such proposals, they will after termination of hostilities take all measures in their power to render impossible the repetition of aggression and violation of peace by Germany or any of the States associated with her in acts of aggression in Europe.
ARTICLE IV
Should either of the high contracting parties during the postwar period become involved in hostilities with Germany or any of the States mentioned in Article III, Section 2, in consequence of the attack by that State against that party, the other high contracting party will at once give to the contracting party so involved in hostilities all military and other support and assistance in his power.
241
This article shall remain in force until the high contracting parties, by mutual agreement, shall recognize that it is superseded by adoption of proposals contemplated in Article III, Section 1. In default of adoption of such proposals, it shall remain in force for a period of twenty years and thereafter until terminated by either high contracting party as provided in Article VIII.
ARTICLE V
The high contracting parties, having regard to the interests of security of each of them, agree to work together in close and friendly collaboration after re-establishment of peace for the organization of security and economic prosperity in Europe.
They will take into account the interests of the United Nations in these objects and they will act in accordance with the two principles of not seeking territorial aggrandizement for themselves and of non-interference in the internal affairs of other States.
ARTICLE VI
The high contracting parties agree to render one another all possible economic assistance after the war.
ARTICLE VII
Each contracting party undertakes not to conclude any alliance and not to take part in any coalition directed against the other high contracting party.
ARTICLE VIII
The present treaty is subject to ratification in the shortest possible time and instruments of ratification shall be exchanged in Moscow as soon as possible.
It comes into force immediately on the exchange of instruments of ratification and shall thereupon replace the agreement between the Government of the Union of Soviet Socialist Republics and His Majesty's Government in the United Kingdom signed at Moscow July 12, 1941.
Part One of the present treaty shall remain in force until the re-establishment of peace between the high contracting parties and Germany and the powers associated with her in acts of aggression in Europe.
Part Two of the present treaty shall remain in force for a period of twenty years. Thereafter, unless twelve months' notice has been given by either party to terminate the treaty at the end of the said period of twenty years, it shall continue in force until twelve months after either high contracting party shall have given notice to the other in writing of his intention to terminate it.
242
In witness whereof the above-named plenipotentiaries have signed the present treaty and have affixed thereto their seals.
Done in duplicate in London on the twenty-sixth day of May, 1942, in the Russian and English language, both texts being equally authentic.
243
Appendix III: The Travis-Wenger Agreement of 1942246
No. OP-20-G/jac 0353520 WASHINGTON October 2, 1942
MEMORANDUM FOR COMMANDER E. W. R. TRAVIS. R.N.:
Subject: Collaboration of U.S. and British radio intelligence organizations on Japanese and German projects.
As a result of discussions of U.S. and British intelligence problems in the Pacific and Atlantic Areas, it is understood that you propose the following, with approval of the Admiralty:
JAPANESE
(a) The British to abandon naval cryptanalysis at Kilindini and retain there only an exploitation unit which will read traffic from recoveries supplied by other units, and supply to these other units any code or other recoveries obtained in the course of this reading. (b) The British to disband the British-Australian naval unit at Melbourne and turn over to the U.S. unit there such personnel as the U.S. may desire, except Commander Nave, who is to be recalled. Requests by the U.S. for any particular individuals from Kilindini or Melbourne will be entertained by the British. The future status of the diplomatic party at Melbourne will depend upon wishes of the Australian Government and the senior naval and military authorities in that area, which the Admiralty will ascertain. (c) Upon execution of the foregoing, OPNAV to assume responsibility for naval recoveries and pertinent naval information to the Admiralty (G.C. & C.S.) for transmittal to C. in C. Eastern Fleet and Kilindini. (d) Pursuant to (c) above, OPNAV to pass to the Admiralty (G.C. & C.S.) (1) radio intelligence from Japanese naval communications, indicating, major strategic moves in any area and any details bearing upon operations in the Indian Ocean Area; (2) All Japanese naval code and cipher key recoveries. (e) In addition to the foregoing, OPNAV to pass to G.C. & C.S. by pouch all Japanese intercepted raw naval traffic.
GERMAN
(a) The British to provide technical assistance, if desired, in the development of analytical machinery required. (b) The British agree in principle to full collaboration upon the German submarine and naval cryptanalysis problems, including exchange of intercepted traffic, keys, menus, cribs, and such other pertinent technical information as may be necessary.
246 Erskine 1999, 192-197.
244
MISCELLANEOUS
(a) The U.S. to undertake certain work on Italian naval systems; traffic, and such pertinent information as may be available to be supplied by the British. (b) British to obtain certain items of special analytical equipment developed by the U.S. (c) The British to send certain technical personnel to OP-20-G to obtain information concerning new U.S. high-speed analytical equipment and the technique employed in certain phases of U.S. work. (d) Direct cable communications to be provided by the U.S, and British between G.C. & C.S. and OP-20-G with each party making its own terminal arrangements.
The results of the foregoing will be that the British will withdraw from active cryptanalytical work in the Pacific Area but will continue to intercept and read Japanese traffic at Kilindini. To the U.S. will be left the general direction and control of the effort against Japanese communications. The British plan, however, to maintain a research and intelligence unit at G.C. & C.S. so as not to lose touch with the Japanese problem. With regard to German communications, the British accede to U.S. desires to attack the submarine and naval problems.
The primary concern of the British over U.S. entry into the German field is on the question of security. The British treat German Enigma matters on a higher plane than any other which they handle. They believe that the situation with regard to German communications is quite different from that which is found in the Japanese in that the ramifications of the major cryptographic system used are very great and any disclosures made might defeat the entire effort in every field. The British consider that, in going into the work, the U.S. should be prepared to accept British standards of security and insure compliance therewith.
It would appear that the foregoing proposals if accepted, would result in a logical set- up for the reason that the U.S. has the primary facilities and experience in the Pacific and is in a position to intercept traffic there with greater success; whereas, the British occupy the corresponding position in the Atlantic. While providing for a logical division of labor on this basis, it would appear, moreover, that this plan will, at the same time, provide a necessary back-up for the safety of each party concerned.
Intelligence, communications, and operational authorities concerned in the Navy Department have considered these proposals and find them acceptable, subject to the following reservations: (a) As for security, the U.S. feels great concern over the treatment of intelligence obtained from Japanese naval communications and desires that it be handled on the same plane as the German. It is felt that there should be a definite agreement as to the dissemination to be given any recoveries and information supplied; and that the instructions for handling this material be as uniform as possible. (b) The extent to which information, recoveries and raw material can be supplied will naturally depend upon communications and other facilities available. (c) The proposals concerning the Melbourne unit have been referred to the Commander, Southwest Pacific Force, for comment, and decision thereon is withheld pending his reply. (d) Any agreement made at this time must be subject to such change as circumstances and developments require. Should it become necessary to make
245
any change, effort will be made to notify you in advance, or, if this is not possible, at the earliest practicable time, thereafter. (e) The question of supplying special analytical equipment will have to be investigated, due to the production problems involved and the possibility of material shortages. A memorandum on this matter will be sent to the British at a later date.
C. F. HOLDEN Captain U.S. Navy Director of Naval Communications
246
Appendix IV: The BRUSA Agreement of 1943247
May 17, 1943
Agreement between British Government Code and Cipher School and U.S. War Department concerning cooperation in matters relating to:
U.S. British Special Intelligence A Special Intelligence Special Intelligence B Y Intelligence TA Intelligence Y Inference
A distinction is made in nomenclature and procedure in handling intelligence derived from the solution of enemy high grade and that obtained from low grade codes and ciphers. The preservation of secrecy in regard to either category is a matter of great concern to both countries and if the highest degree of security is to be maintained, it is essential that the same methods should be pursued by both countries at every level and in every area concerned, since a leakage at any one point would jeopardize intelligence from these sources not in one area only but in all theaters of war and for all services.
This agreement is limited to the traffic specifically designated herein. It does not cover traffic emanating from non-service enemy or neutral sources. These subjects will be covered by future negotiations between Director, G.C.C.S. and A.C. of S., G-2, War Department.
(1) Both the U.S. and British agree to exchange completely all information concerning the detection, identification and interception of signals from, and the solutions of codes and ciphers used by, the Military and Air forces of the Axis powers, including secret services (Abwehr). (2) The U.S. will assume as a main responsibility the reading of Japanese Military and Air codes and ciphers. (3) The British will assume as a main responsibility the reading of German and Italian Military and Air codes and ciphers. (4) Both countries agree that special security regulations shall apply to Intelligence obtained from decoding telegrams in enemy high grade codes and ciphers. (5) Both countries agree to use their most secure codes and ciphers for transmission of the decodes of enemy signals and transmission of technical cryptanalytic data. (6) British or U.S. Commanders-in-Chief, Military or Air, will receive all Special Intelligence necessary to them for the conduct of their operations from either British or U.S. centers as may be mutually agreed. Liaison officers will be
247 “Agreement between British Government Code and Cipher School and U.S. War Department in regard to certain ‘Special Intelligence.’” June 15, 1943.
247
appointed as desired for facilitating this. They will be given full access to all decodes. (7) The distribution of intelligence form the sources in question will be governed by the fundamental principle that distribution will be restricted to the minimum and will therefore be confined solely to those who require to receive the intelligence for the proper discharge of their duties. (8) All recipients of Special Intelligence A, whether British or American officers, shall be bound by the same regulations, the regulations (Appendix B) now in force in the theaters of war where British forces are operating to be accepted at the present time. If at a later date either country wishes to modify them in the light of further experience then this may be done by mutual agreement. (9) The extension to officers of a knowledge of the existence of such intelligence shall be confined to as limited a number as possible and restricted to the levels of command in conformity with the above mentioned regulations. Great stress is laid on the principle that Special Intelligence A should not be intermingled in reports with general intelligence from other sources. If, however, it becomes imperative to do so, the whole must be treated as Special Intelligence A and given the same strictly limited distribution. Under no circumstances is it permissible to pass Special Intelligence A in a code or cipher which can be read by other than the authorized recipients. (10) Although Special Intelligence B is not subject to the same stringent regulations as Special Intelligence A, since the two are closely connected, it is essential to maintain a high degree of secrecy in the handling of Special Intelligence B also. In any action taken upon such intelligence and in any documents or telegrams based upon it, it is essential that its origin be disguised and that the codes or ciphers used for its dissemination be absolutely secure. (11) All intelligence available from decodes shall be made available to Liaison officers, and if they deem necessary it will be exchanged between London and Washington. These Liaison officers will be specifically appointed and given full facilities for this purpose. (12) British and U.S. will notify one another without delay, giving full particular, when either has information from any source indicating the compromise of any code or cipher used by the other. Action on such information will be most carefully considered in order not to compromise the source and if possible mutual agreement in such action will be sought. (13) Cooperation between and coordination of U.S. Signal Intelligence Service and British ‘Y’ Service must take place at all levels, technical information being exchanged mutually at the same level and each country to agree not to lower the classification of such information or the intelligence derived from it below that level without mutual agreement. (14) Each country shall inform the other of the employment and scope in each joint theater of war of their Signal Intelligence (Y) units in the field. (15) This agreement or the appendices thereto may be supplemented or modified from time to time governing any special feature for which either party wishes to make special provision. (16) Definitions:
248
a. Y Service or Signal Intelligence Service. The British, U.S. Army, and U.S. Navy services concerned with intercepting, decoding, intercepting, classifying and dissemination of enemy (and neutral) communications, and the use of D/F and other specialized apparatus for establishing locations and identities of enemy transmitters. b. Special Intelligence A. Certain ciphers are placed in a special category, owing to their importance and difficulty of solution. The intelligence derived from these ciphers is known as Special Intelligence A. Such material is treated with most stringent security measures. Special Intelligence A is confined to a very strictly limited number of the most highly placed officers and is mainly of strategical [sic] importance. c. Special Intelligence B. Intelligence derived from the solution of lower grade ciphers. Such ciphers may under certain circumstances be upgraded to the “Special A” class. The dissemination of Special Intelligence B is wider though always treated as British Most Secret— U.S. Secret. Special Intelligence B may be used tactically.
Appendix (A) Special Provisions Regarding Work on German Machine Ciphers
Since it is believed unnecessary and impracticable to duplicate work on German machine ciphers and in view of the large number of personnel required and the unavoidable extra risk to the security of the source involved, agreement which follows has been arrived at. This agreement provides that:
a. All desired intelligence from this source will be made available to the War Department in Washington. b. U.S. personnel will obtain experience by engaging in the solution of this type of cipher in Great Britain. c. Research into new methods of attack will be made in Washington. d. Transmission of Intelligence to Commanders-in-Chief in the field will be accomplished by special routes and staffs who will maintain a watch over the use of the intelligence to guard against compromise of the source. 1. U.S. liaison officers will be appointed at G. C. & C.S. to examine messages and summaries and select those desired for transmittal to Washington for G-2 or the Theater Commanders. All decoded material will be made available to those officers. Decodes giving information regarding Order of Battle will be handled as at present, i.e., through U.S. liaison officers in War Office and Air Ministry, respectively. 2. Decodes or summaries to be passed to Washington through existing British channels. 3. U.S. party to effect independent solution of keys will be established in Great Britain, but so coordinated by mutual agreement to avoid duplication. This party will cooperate with the British in regard to
249
tasks and will be given every assistance for instruction of personnel. They will be furnished British machines. Decodes from this section will be passed to Bletchley Park for emendation, translation and distribution, but U.S. party will conduct complete processing, including emendation and translation to such an extent as they desire. 4. Formulas will be supplied by Great Britain for use on machines now at Arlington Hall. 5. U.S. to undertake research for finding a new method for solution and to be rendered every assistance by the British for this purpose. 6. In conformity with British policy, U.S. personnel engaged in solution work in Great Britain will not be transferred elsewhere except for very urgent reasons. 7. Special Intelligence from this source will be passed to Commanders- in-Chief in the field through the Special British units provided for this purpose. The officer in command of these units will have direct access to the Commander-in-Chief and advise as necessary on the security aspect of handling and using this intelligence. Where an American officer s Commander-in-Chief, an American officer, properly trained and indoctrinated at Bletchly [sic] Park, will be attached to the unit to advise and act as liaison officer to overcome difficulties that may arise in regard to differences in language. 8. The Director of the G. C. & C. S. will have the final decision when matters of security are involved in intelligence items (gossip) and as to what is passed to Commanders-in-Chief in the field.
Appendix (B) British Security Regulations for Special Intelligence CO-ORDINATION OF ROUTING, SECURITY AND USE OF SPECIAL INTELLIGENCE
Part I 1) SPECIAL INTELLIGENCE is the agreed name for the highly secret information obtained by cryptographic means from enemy high grade ciphers. 2) Lower grade cryptographic material classed in general as “Y” Intelligence, is not included in the definition of SPECIAL INTELLIGENCE.
PART I TO BE DESTROYED BY FIRE WHEN READ
Part II 1) ALL SPECIAL INTELLIGENCE emanating from the United Kingdom and transmitted to Commands abroad will receive the prefix ‘Ultra’. 2) ALL SPECIAL INTELLIGENCE emanating from centres other than the United Kingdom and transmitted either to the United Kingdom or to another Command abroad, is to receive the prefix specially allotted to each producing centre as follows: -
250
DELHI Prefix SIRDAR WASHINGTON Prefix ZYMOTIC MELBOURNE Prefix ZYMOTIC KILINDINI Prefix ZYMOTIC MIDDLE EAST Prefix SWELL
3) SPECIAL INTELLIGENCE produced by U.S.A. centres either in U.S.A. or elsewhere if transmitted over British routes either to the United Kingdom or to British Commands overseas, is to receive the prefix of the Command or centre through which it is distributed. 4) Where it is necessary for SPECIAL INTELLIGENCE to be transmitted between Commands or centres other than the United Kingdom, special routes and ciphers are to be arranged and approved by London.
Part III
ULTRA (see Para. 1 above) information can be regarded as reliable and action can be taken on it, but experience has shown that the following security regulations are vital to the preservation of this source. The Commander-in-Chief is held personally responsible for ensuring that they are scrupulously adhered to: -
1) The utmost secrecy is to be used in dealing with ULTRA information. Attention is called to the fact that if from any document that might fall into the hands of the enemy or from any message that the enemy might intercept, from any word that might be revealed by a prisoner of war, or from any ill- considered action based upon it, the enemy were to suspect the existence of the ULTRA source, that source would probably forever be lost to our cause. 2) This loss would vitally affect operations on all fronts, not only the particular front on which the source had been compromised. 3) Commanding Officers of those Commands authorized to receive ULTRA information, i.e., normally only General and Air Officers commanding Armies and Air Forces, are to be instructed that ULTRA messages are for them, their personal representative, and their Senior Intelligence and Operations Staff Officer only, and are not to be seen by, read to, or discussed with any other person. ULTRA messages are to be destroyed by fire immediately [sic] action has been taken on them. No record of Intelligence based on ULTRA information may be kept, except at the H.Q. of the Commander-in-Chief. 4) When ULTRA information is to be used by the Commander of an Army or an Air Force as a basis for action to be taken by a subordinate command, the information must be translated, when passed to the subordinate command, into terms of an operational order, so worded that if captured or intercepted by the enemy the origin of the information could not be traced back to the ULTRA source, e.g., orders must never contain the precise time, date or place of an enemy operation revealed by ULTRA. Such orders based on ULTRA
251
information if transmitted by W/T must be encoded only in authorized ciphers. Under no circumstances whatever is it permissible to transmit ULTRA information as such to lower formations. 5) In general, if any action is to be taken based upon ULTRA information, the local Commander is to ensure that such action cannot be traced back by the enemy to the reception of ULTRA Intelligence alone. A momentary tactical advantage is not sufficient ground for taking any risk of compromising the source. No action may be taken against specific sea or land targets revealed by ULTRA unless appropriate air or land reconnaissance has also been undertaken. Names of enemy ships revealed by ULTRA source may never be quoted. 6) The utmost care is to be taken in briefing pilots for an operation based on ULTRA information that only such details are given them as might have been obtained by other means, such as air reconnaissance, and only such as are essential to the success of the operations. 7) No reference to ULTRA information is to be made in any summary whatsoever, however limited the circulation. No discussion of it is permissible except between the senior officers who are immediately concerned with the action to be taken upon it. 8) If it is necessary to ask questions, or make comments on ULTRA material, whether on matters of Intelligence, Operations, Routing or Security, such messages are to be transmitted only over the special channel and in the special ciphers provided for ULTRA traffic. 9) Recipients of ULTRA may not under any circumstances carry on their persons outside their Headquarters, ULTRA messages which have been delivered to them.
1st March 1943
(Signed) GEO. V. STRONG Major General A.C. of S., G-2
(Signed) E. W. TRAVIS D.D.(S) G.C.&C.S.
15 June 1943
Approved for the U.S. War Department. By order of the Secretary of War. (Signed) JOSEPH T. McNARNEY Lieut. General, U.S. Army Deputy Chief of Staff
252
Bibliography
PRIMARY SOURCE DOCUMENTS
The National Archives (Kew, England)
CAB 24. War Cabinet and Cabinet: Memoranda (GT,CP and G War Series). 1915- 1939.
CAB 122. War Cabinet and Cabinet Office: British Joint Staff Mission and British Joint Services Mission: Washington Office Records. 1940-1958.
HW 8. Government Code and Cypher School (GCCS): Naval Section: Reports, Working Aids and Correspondence. 1914-1946.
HW 14. GCCS: Directorate: Second World War Policy Papers. 1933-1949.
HW 16. GCCS: Government Code and Cypher School: German Police Section: Decrypts of German Police Communications during Second World War. 1939-1945.
HW 50. GCCS and Government Communications Headquarters (GCHQ): Records relating to the writing of the history of British signals intelligence in World War II. 1914-1954.
HW 57. GCCS: Liaison with Signals Intelligence Organisations in the United States of America. 1941-1946.
HW 61. GCCS: Recods relating to Commonwealth and Allied Signals Intelligence Organisations. 1943-1945.
HW 80. General Records of GCHQ. 1945-1956.
KV 4/447. Organisation of and Liaison with British Security Coordination, USA. 1942-1945.
WO 178. War Office: British Military Missions: War Diaries, Second World War. 1939-1946.
WO 208. War Office: Directorate of Military Operations and Intelligence, and Directorate of Military Intelligence; Ministry of Defence, Defence Intelligence Staff : Files. 1917-1974.
National Archives and Records Administration (College Park, Maryland)
RG 457. Records of the National Security Agency/Central Security Service [NSA/CSS]. 1917-1993.
253
BOOKS AND ARTICLES
Abbott, Kenneth W., and Duncan Snidal. “Why States Act through Formal International Organizations.” The Journal of Conflict Resolution 42, no. 1 (1998): 3-32.
Abrams, Philip. "Notes on the Difficulty of Studying the State." Journal of Historical Sociology 1, no. 1 (1988 (1977)): 58-89.
Aid, Matthew M. "All Glory is Fleeting: Sigint and the Fight Against International Terrorism." Intelligence and National Security 18, no. 4 (2003): 72-120.
Aid, Matthew M. "The National Security Agency and the Cold War." Intelligence and National Security 16, no. 1 (2001a): 27-66.
Aid, Matthew M., and Cees Wiebes. "Introduction: The Importance of Signals Intelligence in the Cold War." Intelligence and National Security 16, no. 1 (2001b): 1-26.
Aldrich, Richard J. "British Intelligence and the Anglo-American 'Special Relationship' during the Cold War." Review of International Studies 24, no. 1 (1998): 331-351.
Aldrich, Richard J. "GCHQ and Sigint in Early Cold War, 1945-70." Intelligence and National Security 16, no. 1 (2001): 67-96.
Aldrich, Richard J. "Dangerous Liaisons: Post-September 11 Intelligence Alliances." Harvard International Review 24, no. 3 (2002a).
Aldrich, Richard J. "'Grow Your Own': Cold War Intelligence and History Supermarkets." Intelligence and National Security 17, no. 1 (2002b): 135-152.
Aldrich, Richard J. "Never-Never Land or Wonderland? British and American Policy on Intelligence Archives." Studies in Intelligence (Central Intelligence Agency) 38, no. 5 (1995): 17-26.
Aldrich, Richard J. "Transatlantic Intelligence and Security Cooperation." International Affairs 80, no. 3 (2004): 733-755.
Alvarez, David, ed. Allied and Signals Intelligence in World War II. Portland, OR: Frank Cass, 1999.
Alvarez, David, ed. Secret Messages: Codebreaking and American Diplomacy, 1930- 1945. Lawrence, KS: University of Kansas Press, 2000.
254
Amplitude. Dictionary.com. Dictionary.com Unabridged. Random House, Inc. http://dictionary.reference.com/browse/amplitude (accessed June 25, 2012).
Anderson, Elizabeth. "The Security Dilemma and Covert Action: The Truman Years." International Journal of Intelligence and Counterintelligence 11, no. 4 (1998/99): 403-427.
Andrew, Christopher. For the President's Eyes Only: Secret Intelligence and the American Presidency from Washington to Bush. NY, NY: HarperPerennial, 1995.
Andrew, Christopher. "Intelligence analysis needs to look backwards before looking forward." History and Policy. June 1, 2004a. http://www.historyandpolicy.org/papers/policy-paper-23.html (accessed April 4, 2008).
Andrew, Christopher. "Intelligence, International Relations and 'Under-theorisation'." Intelligence and National Security 19, no. 2 (2004b): 170-184.
Andrew, Christopher, and David Dilks, . The Missing Dimension: Governments and Intelligence Communities in the Twentieth Century. London: University of Illinois Press, 1984.
Andrew, Christopher, and Vasili Mitrokhin. The Sword and the Shield: The Mitrokhin Archive and the Secret History of the KGB. NY, NY: Basic Books, 1999.
Armitage, Richard. "Intelligence Policy and National Policy Coordination." Panel IV of Day Two of the Eighth Public Hearing of the National Commission on Terrorist Attacks Upon the United States. March 24, 2004.
The Avalon Project. Twenty-Year Mutual Assistance Agreement Between the United Kingdom and the Union of Soviet Socialist Republics: May 26, 1942. http://avalon.law.yale.edu/wwii/brsov42.asp (accessed July 5, 2012).
Aydinli, Ersel, and Musa Tuzuner. “Quantifying Intelligence Cooperation: The United States International Intelligence Behavior (USIIB) Dataset.” Journal of Peace Research 48, no. 5 (2011): 673-682.
Babst, Dean V. "A Force for Peace." Industrial Research, April 1972: 55-58.
Babst, Dean V. "Elective Governments - A Force for Peace." The Wisconsin Sociologist 3, no. 1 (1964): 9-14.
Bailey, Roderick. “OSS-SOE Relations, Albania 1943-44.” In American-British- Canadian Intelligence Relations 1939-2000, edited by David Stafford and Rhodri Jeffreys-Jones, 20-35. London: Frank Cass, 2000.
255
Bates, Robert H. "Area Studies and the Discipline: A Useful Controversy?" PS: Political Science and Politics 30, no. 2 (1997): 166-169.
Battersby, John. "How a Cycle of Violence Entraps Arafat in Peace Talks." Christian Science Monitor, January 31, 1996: 1.
Beardsley, E.H. "Secrets between Friends: Applied Science Exchange between the Western Allies and the Soviet Union during World War II." Social Studies of Science 7, no. 4 (November 1977): 447-473.
Beer, Siegfried. "Intelligence Institutions and State Relations in the Twentieth Century: a Central European Perspective." In Public Power in Europe: Studies in Historical Transformations, edited by James S Amelang and Siegfried Beer, 181-205. Pisa: Pisa University Press, 2006.
Bell, Christopher M. "Winston Churchill and the Ten Year Rule." The Journal of Military History 74, no. 1097 (October 2010): 1128.
Benson, Robert Louis. “A History of U.S. Communications Intelligence during World War II: Policy and Administration.” United States Cryptologic History Series IV, Volume 8 (1997). http://www.nsa.gov/about/_files/cryptologic_heritage/publications/wwii/histor y_us_comms.pdf (accessed June 24, 2012).
Benson, Robert Louis. “The Origin of U.S.-British Communications Intelligence Cooperation (1940-41). Cryptologic Spectrum (National Security Agency) 7, no. 4 (1977). www.nsa.gov/public_info/_files/cryptologic_spectrum/origin_us_british.pdf (accessed June 24, 2012).
Bennett, Andrew, and Colin Elman. "Qualitative Research: Recent Developments in Case Study Method." Annual Review of Political Science 9 (2006): 455-476.
Bennett, Gill. “A Most Extraodinary and Mysterious Business: The Zinoviev Letter of 1924.” Foreign & Commonwealth Office (FCO) History Note No. 14. London: FCO, 1999. http://www.fco.gov.uk/resources/en/pdf/pdf5/fco_pdf_zinovievletter2 (accessed Novemver 17, 2009).
Bensahel, Nora. "A Coalition of Coalitions: International Cooperation Against Terrorism." Studies in Conflict and Terrorism 29, no. 1 (2006): 35-49.
Berkowitz, Bruce. "Intelligence Reform, The Third Act." Orbis 45, no. 1 (2001): 152- 162.
Best, R.A. Cooperation with Like-Minded Peoples: British Influences on American Security Policy, 1945-1949. Westport, CT: Greenwood, 1986.
256
Betts, Richard. Enemies of Intelligence: Knowledge and Power in American National Security. NY, NY: Columbia University Press, 2007.
Betts, Richard. "The Soft Underbelly of American Primacy: Tactical Advantages of Terror." Political Science Quarterly 117, no. 1 (2002): 19-36.
Boatner, Helene L. "Sharing and Using Intelligence in International Organizations: Some Guidelines." National Security and the Future 1, no. 1 (2000): 81-92.
Bonsall, Arthur. "Bletchley Park and the RAF Y Service: Some Recollections." Intelligence and National Security 23, no. 6 (2008): 827-841.
Brady, Henry E., David Collier, and Jason Seawright. "Refocusing the Discussion on Methodology." In Rethinking Social Inquiry: Diverse Tools, Shared Standards, by Henry E. Brady and David Collier, 3-20. Lanham, MD: Rowman and Littlefield, 2004.
Bratton, Michael, and Nicholas van de Walle. Democratic Experiments in Africa: Regime Transitions in Comparative Perspective. NY, NY: Cambridge University Press, 1997.
Brooks, Stephen G. "Dueling Realisms." International Organization 51, no. 3 (1997): 445-477.
Bruneau, Thomas C., and Kenneth R. Dombroski. "Reforming Intelligence: The Challenge of Control in New Democracies." In Who Guards the Guardians and How: Democratic Civil-Military Relations, edited by Thomas C. Bruneau and Scott D. Tollefson, 145-177. Austin, TX: University of Texas Press, 2006.
Bruneau, Thomas C., and Scott D. Tollefson, . Who Guards the Guardians and How: Democratic Civil-Military Relations. Austin, TX: University of Texas Press, 2006.
Bruneau, Thomas C., and Stephen C. Boraz, . Reforming Intelligence: Obstacles to Democratic Control and Effectiveness. Austin, TX: University of Texas Press, 2007.
Budiansky, Stephen. "The Difficult Beginnings of US-British Codebreaking Cooperation." Intelligence and National Security 15, no. 2 (2000): 49-73.
Bueno de Mesquita, Bruce, James D. Siverson, Randolph M. Morrow, and Alastair Smith. "An Institutional Explanation of the Democratic Peace." American Political Science Review 93, no. 4 (1999): 791-807.
Byman, Daniel. "Prepared Statement of Dr. Daniel Byman." Extraordinary Rendition, Extraterritorial Detention, and Treatment of Detainees: Restoring Our Moral Credibility and Strengthening Our Diplomatic Standing. Washington, DC, July 26, 2007.
257
Byman, Daniel. "Remaking Alliances for the War on Terrorism." The Journal of Strategic Studies 29, no. 5 (2006): 767-811.
Cahn, Anne H. Killing Detente: The Right Attacks the CIA. University Park, PA: The Pennsylvania State University Press, 1998.
Cammack, Paul. "Bringing the State Back In: A Polemic." British Journal of Political Science 19, no. 2 (1989): 261-290.
Carothers, Thomas. "Promoting Democracy and Fighting Terror." Foreign Affairs 82, no. 1 (2003).
Carothers, Thomas. "The Democracy Crusade Myth." The National Intereset Online, June 2007.
Carothers, Thomas. "The End of the Transition Paradigm." Journal of Democracy 13, no. 1 (2002): 8-9.
CBSNews. “Faulty Intel Source ‘Curve Ball’ Revealed.” February 11, 2009. cbsnews.com, February 11, 2009. http://www.cbsnews.com/2100-18560_162- 3440577.html (accessed July 10, 2012).
Chan, Steve. "In Search of Democratic Peace: Problems and Promise." Mershon International Studies Review 41, no. 5 (1997): 59-91.
Charles, Douglas M. "American, British and Canadian Intelligence Links: A Critical Annotated Bibliography." Intelligence and National Security 15, no. 2 (2000): 258-269.
Charles, Douglas M. "‘Before the Colonel Arrived’: Hoover, Donovan, Roosevelt, and the Origins of American Central Intelligence, 1940–41." Intelligence and National Security 20, no. 2 (2005): 225-237.
Chernoff, Fred. "The Study of Democratic Peace and Progress in International Relations." International Studies Review 6, no. 1 (2004): 49-77.
Choi, Ajin. "The Power of Democratic Cooperation." International Security 28, no. 1 (2003): 142-153.
Christensen, Thomas, and Jack Snyder. "Chain Gangs and Passed Bucks: Predicting Alliance Patterns in Multipolarity." International Organization 44, no. 2 (1990): 137-168.
Clift, Denis. "Through the Prism of National Security: The Challenge of Intelligence Sharing." Defense Intelligence Journal 11, no. 1 (2002): 97-104.
258
Clough, Christopher. "Quid Pro Quo: The Challenges of International Strategic Intelligence Cooperation." International Journal of Intelligence and Counterintelligence 17, no. 4 (2004): 601-613.
Coase, R.H. “The Nature of the Firm.” Economica 4, no. 16 (1937): 386-405.
Collier, David. "Glossary." In The New Authoritarianism in Latin America, edited by David Collier, 399+. Princeton, NJ: Princeton University Press, 1980.
Collier, David, and Steven Levitsky. "Democracy with Adjectives: Conceptual Innovation in Comparative Research." World Politics 49, no. 3 (1997): 430- 451.
Continuum. Merriam-Webster Online. 2012. www.merriam-webster.com (accessed June 25, 2012).
Correlates of War Project. Direct Contiguity Data, 1816-2006. Version 3.1 (2007). http://correlatesofwar.org (accessed March 2007).
Dallek, Robert. Franklin D. Roosevelt and American Foreign Policy, 1932–1945. Oxford: Oxford University Press, 1995.
Danchev, Alex. "On Friendship: Anglo-America at fin de siecle." International Affairs 73, no. 4 (1997): 747-759.
Danchev, Alex. "On Specialness." International Affairs (Royal Institute of International Affairs) 72, no. 4 (October 1996): 737-750.
Davies, Phillip. "Spies as informants: Triangulation and the Integration of Elite Interview Data in the Study of the Intelligence and Security Services." Politics, 2001: 73-80.
Davis, Jack. “Improving Intelligence Analysis at CIA: Dick Huer’s Contribution to Intelligence Analysis.” Center for the Study of Intelligence (Central Intelligence Agency) (1999). http://www.au.af.mil/au/awc/awcgate/psych- intel/art3.html (accessed 24 June 2012).
Defense Committee of the Assembly of Western European Union. "The New Challenges Facing European Intelligence." Assembly of Western European Union, The Interim European Security and Defense Assembly, 48th session. June 4, 2002. http://www.assembly- weu.org/en/documents/sessions_ordinaires/rpt/2002/1775.pdf?PHPSESSID=e 5b536a7a937527d03fbdc8a794e4baf (accessed March 27, 2008).
Dilanian, Ken and Brian Bennett. “Al Qaeda bomb plot was foiled by double agent.” latimes.com, May 9, 2012. http://articles.latimes.com/2012/may/09/world/la- fg-bomb-plot-20120509 (accessed June 26, 2012).
259
Dixon, William. "Democracy and Peaceful Settlement of International Conflict." American Political Science Review 88, no. 3 (1994): 14-32.
Doyle, Michael W. "To the Editors (Michael Doyle on the Democratic Peace)." International Security 19, no. 2 (1995): 180-184.
Doyle, Michael W. "Liberalism and World Politics." American Political Science Review 80, no. 4 (1986): 1151-1169.
Drogin, Bob and John Goetz. “How U.S. Fell Under the Spell of ‘Curveball.’” latimes.com, November 20, 2005. http://www.latimes.com/news/nationworld/iraq/complete/la-na- curveball20nov20,0,5362808,full.story (accessed July 10, 2012).
Druckman, Daniel. Doing Research: Methods of Inquiry for Conflict Anlaysis. Thousand Oaks, CA: Sage Publications, 2005.
Durbin, Brent. "Reforming U.S. Intelligence: When and How Should Intelligence Be Centralized?" All Academic, Inc. March 26, 2008. http://www.allacademic.com/meta/p2522530_index.html (accessed April 3, 2008).
Eckstein, Harry. "Case Study and Theory in Political Science." In Regarding Politics: Essays on Political Theory, Stability, and Change, by Harry Eckstein, 117- 228. Berkeley, CA: University of California Press, 1992 (1975).
Ermarth, Fritz. "Oleg Penkovsky's World." Open Society Archives (OSA). 11 22, 1965. http://www.osaarchivum.org/files/holdings/300/8/3/text/63-1-76.shtml (accessed September 7, 2010).
Erskine, Ralph. "William Friedman's Bletchley Park Diary: A Different View." Journal of Intelligence and National Security 22, no. 3 (2007): 367-379.
Erskine, Ralph. "The Holden Agreement on Naval Sigint: The First BRUSA?” Intelligence and National Security 14, no. 2 (1999): 187-197.
Erskine, Ralph. “The 1944 Naval BRUSA Agreement and its Aftermath.” Cryptologia 30, no. 1 (2006): 1-22.
Fearon, James. "Domestic Political Audiences and the Escalation of International Disputes." The American Political Science Review 88, no. 3 (1994): 577-592.
Ferris, John R. ""Now that the Milk Is Spilt": Appeasement and the Archive on Intelligence." Diplomacy and Statecraft 19 (2008): 527-565.
Ferris, John R. "Treasury Control, the Ten Year Rule and British Service Policies, 1919-1924." The Historical Journal 30, no. 4 (1987): 859-883.
260
Florence, Gregory J. Courting a Reluctant Ally: An Evaluation of U.S./UK Naval Intelligence Cooperation, 1935-1941. Washington, DC: Center for Intelligence Research, Joint Military Intelligence College, 2004.
Flory, Harriette. "The Arcos Raid and the Rupture of Anglo-Soviet Relations, 1927." Journal of Contemporary History 12, no. 4 (1977): 707-723.
Frieden, Jeffrey A. "Actors and Preferences in International Relations." In Strategic Choice and International Relations, edited by David A. Lake and Robert Powell, 39-76. Princeton, NJ: Princeton University Press, 1999.
Fry, M.G., and M. Hochstein. "Epistemic Communities: Intelligence Studies and International Relations." In Espionage: Past, Present, Future?, edited by Wesley K. Wark, 14-28. Portland, OR: Frank Cass, 1994.
Gartzke, Erik. "Kant We All Just Get Along? Opportunity, Willingness, and the Origins of the Democratic Peace." American Journal of Political Science 42, no. 1 (1998): 1-27.
Geddes, Barbara. “How the Cases You Choose Affect the Answers You Get: Selection Bias in Comparative Politics.” Political Analysis. 2, no. 1 (1990): 131-150.
George, Alexander L., and Andrew Bennett. Case Studies and Theory Develoment in the Social Sciences. Cambridge, MA: MIT Press, 2005.
Gerring, John. "Causal Mechanisms: Yes, But…." Comparative Political Studies 44, no. 5 (2010).
Gerring, John. "Is There a (Viable) Crucial-Case Method?" Comparative Political Studies 40, no. 3 (2007): 231-253.
Gibbs, David N. "Academics and Spies: The Silence that Roars." Los Angeles Times, January 28, 2001.
Gibler, Douglas, and Toby J. Rider. "Prior Commitments: Compatible Interests Versus Capabilities in Alliance Behavior." International Interactions 30 (2004): 309-329.
Gill, Peter, Stephen Marrin and Mark Phythian, editors. Intelligence Theory: Key Questions and Debates. NY: Routledge, 2009.
Gill, Peter, and Mark Phythian. Intelligence in an Insecure World. Cambridge: Polity, 2006.
Gill, Peter, and Mark Phythian. "Issues in the Theorization of Intelligence." International Studies Association. March 2004.
261
http://convention2.allacademic.com/one/isa/isa04/index.php?cmd=isa04 (accessed August 28, 2007).
Gilligan, Michael. “The Transactions Costs Approach to International Institutions.” Unpublished Working Paper (New York University) (2007). http://politics.as.nyu.edu/docs/IO/2601/TransactionsCostsApproach.pdf (accessed June 24, 2012).
Gladwin, Lee A. "Cautious Collaborators: The Struggle for Anglo-American Cryptanalytic Cooperation 1940-43." In Allied and Axis Signals Intelligence in World War II, edited by David Alvarez. Portland, OR: Frank Cass, 1999.
Glenny, M.V. "The Anglo-Soviet Trade Agreement, March 1921." Journal of Contemporary History 5, no. 2 (1970): 63-82.
Gochman, Charles S. “Interstate Metrics: Conceptualizing, Operationalizing, and Measuring the Geographic Proximity of States since the Congress of Vienna." International Interactions 17, no. 1 (1991): 93-112.
Goodman, Michael S. "Studying and Teaching About Intelligence: The Approach in the United Kingdom." Studies in Intelligence (Central Intelligence Agency) 50, no. 2 (2004): 57-65.
Gowa, Joanne. Ballots and Bullets: The Elusive Democratic Peace. Princeton, NJ: Princeton University Press, 1999.
Grieco, Joseph M. "Anarchy and the Limits of Cooperation: A Realist Critique of the Newest Liberal Institutionalism." International Organization 42, no. 3 (1988): 485-507.
Gwertzman, Bernard. "Iraq Gets Reports from U.S. for Use in War with Iran." The New York Times, December 16, 1986: A1+.
Hainmueller, Jens, and Michael J. Hiscox. "Being Specific: Measuring Asset Specificity for Political Economy." Michael J. Hiscox. August 2007. www.people.fas.harvard.edu/~hiscox/HHSpecificity.pdf (accessed September 30, 2010).
Hall, Peter, and David Soskice (eds.) Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. NY: Oxford University Press, 2001.
Harrison, Mark. “The Economics of World War II: An Overview.” In The Economics of World War II: Six Great Powers in International Comparison, edited by Mark Harrison, 1-42. Cambridge: Cambridge Univeristy Press, 1998. www2.warwick.ac.uk/fac/soc/economics/staff/academic/harrison/public/ww2 overview1998.pdf (accessed June 25, 2012).
262
Hasenclever, Andreas, Peter Mayer, and Volker Rittberger. "Interests, Power, Knowledge: The Study of International Regimes." Mershon International Studies Review 40 (1996): 170-228.
Haufler, Hervie. Codebreakers' Victory: How the Allied Cryptographers Won World War II. NY: New American Library, 2003.
Hayden, Michael V. "Opening Statement by Michael V. Hayden Before the Senate Select Committee on Intelligence." May 18, 2006.
Hayden, Michael V. "Transcript of Interview on The Charlie Rose Show." October 24, 2007.
Hershberg, James G. "Their Men in Havana: Anglo-American Intelligence Exchanges and the Cuban Crises, 1961-1962." In American-British-Canadian Intelligence Relations, 1939-2000, edited by David Stafford and Rhodri Jeffreys-Jones, 121-176. Portland, OR: Frank Cass, 2000.
Hibbs, Mark. "Pakistan Says Access Denied to Khan to Avert Threat to Government." Platts Nuclear Fuel, Deember 18, 2006.
Hill, Alexander. “British Lend Lease Aid and the Soviet War Effort.” The Journal of Military History 71, no. 3 (2007): 773-808.
Hinsley, F.H., E.E. Thomas, et al. British Intelligence in the Second World War: Its Influence Upon Operations. 4 volumes. London: Her Majesty’s Stationery Office, 1979-185.
Horovitz, David. "Israeli Arab held as joint crackdown intensifies." The Irish Times, March 7, 1996: 9.
Hulnick, Arthur S. Fixing the Spy Machine: Preparing American Intelligence for the 21st Century. Westport, CT: Praeger, 1999.
Iversen, Torben, and David Soskice. "Two Worlds of Capitalism: Ricardo versus Heckscher-Ohlin." Paper prepared for presentation at the American Political Science Association meetings in San Francisco, August 29-September 3, 2001. September 2001. http://www.fas.harvard.edu/~iversen/PDFfiles/iversen- soskice.PDF (accessed September 16, 2010).
Jackson, Patrick Thaddeus. “Science, Politics, and Swift Boats.” Paper presented at the annual meeting of the American Political Science Association, Hilton Chicago and the Palmer House Hilton, Chicago, IL. September 2004. http://citation.allacademic.com/meta/p_mla_apa_research_citation/0/6/1/1/8/p ages61181/p61181-1.php (accessed June 25, 2012).
263
James, Patrick, and Glenn E. Mitchell II. "Targets of Covert Pressure: The Hidden Victims of the Democratic Peace." International Interactions 21, no. 1 (1995): 85-107.
Jeffreys-Jones, Rhodri. “The Role of British Intelligence in the Mythologies Underpinning the OSS and Early CIA.” In American-British-Canadian Intelligence Relations 1939-2000, edited by David Stafford and Rhodri Jeffreys-Jones, 5-19. Portland, OR: Frank Cass, 2000.
Jervis, Robert. "Hyoptheses on Misperception." World Politics 20, no. 3 (1968): 454- 479.
Jervis, Robert. Perception and Misperception in International Politics. Princeton, NJ: Princeton University Press, 1976.
Jervis, Robert. "Intelligence and Foreign Policy: A Review Essay." International Security 11, no. 3 (1986): 141-161.
Jervis, Robert. "Realism, Neoliberalism, and Cooperation: Understanding the Debate." International Security 24, no. 1 (1999): 42-63.
Johnson, Chalmers. "Preconception vs. Observation, or the Contributions of Rational Choice Theory and Area Studies to Contemporary Political Science." PS: Political Science and Politics 30, no. 2 (1997): 170-174.
Johnson, Loch K. A Season of Inquiry: The Senate Intelligence Investigation. Lexington, KY: University Press of Kentucky, 1985.
Johnson, Loch K. Secret Agencies: US Intelligence in a Hostile World. New Haven, CT: Yale University Press, 1996.
Johnson, Loch K. "Intelligence and the Challenge of Collaborative Government." Intelligence and National Security 13, no. 2 (1998): 177-182.
Johnson, Loch K. "Bricks and Mortar for a Theory of Intelligence." Comparative Strategy 22, no. 1 (2003): 1-28.
Johnson, Loch K. "Harry Howe Ransom and American Intelligence Studies." Intelligence and National Security 22, no. 3 (2007): 402-428.
Jones, Morgan D. Fourteen Powerful Techniques for Problem Solving. NY, NY: Three Rivers Press, 1998.
Jopson, Barney. "A Dismissive Ethiopia Tests the Limits of US Induglence." Financial Times, October 11, 2007, Asia Edition ed.: 9.
Kahn, David. "Edward Bell and His Zimmermann Telegram Memorandums." Intelligence and National Security 14, no. 3 (1999): 143-159.
264
Kant, Immanuel. “To Perpetual Peace: A Philosophical Sketch.” In Perpetual Peace and Other Essays, translated by Ted Humphrey, 107-144. Indianapolis, IN: Hackett Publishing Company, 1983 (1795).
Katz, Barry M. "The Criticism of Arms: The Frankfurt School Goes to War." The Journal of Modern History 59, no. 3 (1987): 439-478.
Keller, William W. The Liberals and J. Edgar Hoover: Rise and Fall of a Domestic Intelligence State. Princeton, NJ: Princeton University Press, 1989.
Kennedy, Greg. "Intelligence and the Blockade, 1914-17: A Study in Administration, Friction and Command." Intelligence and National Security 22, no. 5 (October 2007): 699-721.
Kennedy, Greg. "Strategy and Power: the Royal Navy, the Foreign Office and the Blockade, 1914–1917." Defence Studies 8, no. 2 (2008): 190-206.
Keohane, Robert O. "The Demand for International Regimes." International Organization 36, no. 2 (1982): 325-355.
Keohane, Robert O. "Theory of World Politics: Structural Realism and Beyond." In Neorealism and Its Critics, edited by Robert O. Keohane, 158-203. NY, NY: Columbia University Press, 1986.
Keohane, Robert O., Gary King, and Sidney Verba. Designing Social Inquiry. Princeton, NJ: Princeton University Press, 1994.
Khong, Yuen Foong. Analogies at War: Korea, Munich, Dien Bien Phu, and the Vietnam Decisions of 1965. Princeton, NJ: Princeton University Press, 1992.
Kinsella, David. "No Rest for the Democratic Peace." American Political Science Review, 2005: 453-457.
Knight, Amy. The KGB: Police and Politics in the Soviet Union. Boston, MA: Allen & Unwin, 1988.
Kochavi, Arieh J. "From Cautiousness to Decisiveness: Changes in Soviet Negotiation Tactics Toward the British During World War II." East European Quarterly XXXVI, no. 4 (January 2003): 481-509.
Koremenos, Barbara, Charles Lipson, and Duncan Snidal. "The Rational Design of International Institutions." International Organization 55, no. 4 (2001): 761- 799.
Kovach, Karen. Breaking Codes, Breaking Barriers: the WACs of the Signal Security Agency, World War II. Office of the Chief of Staff, History Office, Fort Belvoir, VA: US Army Intelligence and Security Command, 2007.
265
Kulish, Nicholas. "Terror Battle Puts Denmark on Front Line." The New York Times, September 17, 2007.
Lai, Brian, and Dan Reiter. "Democracy, Political Similarity, and International Alliances, 1816-1992." The Journal of Conflict Resolution 44, no. 2 (2004): 203-227.
Lakatos, Imre. “Falsification and the Methodology of Scientifice Research Programmes.” In Criticism and the Growth of Knowledge, edited by Imre Lakatos and Alan Musgrave, 91-195. NY: Cambridge University Press, 1970.
Lake, David A. "Anarchy, Hierarchy and the Variet of International Relations." International Organization 50, no. 1 (1996): 1-33.
Lake, David A. and Robert Powell. “International Relations: A Strategic-Choice Approach.” In Strategic Choice and International Relations, edited by David A. Lake and Robert Powell, 3-38. Princeton, New Jersey: University of Princeton Press, 1999.
Lake, David A. "Beyond Anarchy: The Importance of Security Institutions." International Security 26, no. 1 (2001): 129-160.
Lake, David A. “The New Sovereignty in International Relations.” International Studies Review no. 5 (2003): 303-323.
Lake, David A. “Escape from the State of Nature: Authority and Hierachy in World Politics.” International Security 32, no. 1 (2007): 47-79.
Landau, Henry. The Enemy Within: The Inside Story of German Sabotage in America. New York: G.P. Putnam's Sons, 1937.
Lander, Sir Stephen. "International Intelligence Cooperation: An Inside Perspective." Cambridge Review of International Affairs 16, no. 4 (2004): 527-542.
Langer, John Daniel. “The Harriman-Beaverbrook Mission and the Debate over Unconditional Aid for the Soviet Union, 1941.” Journal of Contemporary History 14, no. 3 (1979): 463-482.
Larsen, Daniel. "British Intelligence and the 1916 Mediation Mission of Colonel Edward M. House." Intelligence and National Security 25, no. 5 (2010): 682- 704.
Layne, Christopher. "Kant or Can't: The Myth of the Democratic Peace." International Security 19, no. 2 (1994): 5-49.
Leeds, Brett Ashley, and Sezi Anaz. "Alliance Institutionalization and Alliance Performance." International Interactions 31, no. 3 (2005): 183-202.
266
Leeds, Brett Ashley, Andrew G. Long, and Sarah McLaughlin Mitchell. "Reevaluating Alliance Reliability: Sepcific Threats, Specific Promises." The Journal of Conflict Resolution 44, no. 5 (2000): 686-699.
Lefebvre, Stéphane. "The Difficulties and Dilemmas of International Intelligence Cooperation." International Journal of Intelligence and Counterintelligence 16, no. 4 (2003): 527-542.
Legro, Jeffrey W. "Whence American Internationalis ." International Organization 54, no. 2 (2000): 253-289.
Levitsky, Steven, and Lucan A. Way. "Elections Without Democracy: The Rise of Competitive Authoritarianism." Journal of Democracy 13, no. 2 (April 2002): 51-65.
Levy, Jack S. "Domestic Politics and War." In The Origin and Prevention of Major Wars, edited by Robert I. Rotberg and Theodore K. Rabb, 79-100. NY: Cambridge University Press, 1989.
Levy, Jack. "Explaining Events and Developing Theories: History, Political Science, and the Analysis of International Relations." In Bridges and Boundaries: Historians, Political Scientists, and the Study of International Relations, edited by Colin Elman and Miriam Fendius Elman, 39-84. Cambridge, MA: MIT Press, 2001.
Linz, Juan J. "An Authoritarian Regime: Spain." In Cleavages, Ideologies, and Party Systems, edited by Erik Allardt and Yrjo Littunen. Helsinki: Academic Bookstore, 1964.
Little, Douglas. "Red Scare, 1936: Anti-Bolshevism and the Origins of British Non- Intervention in the Spanish Civil War." Journal of Contemporary History 23, no. 2 (1988): 291-311.
Lowenthal, Mark M. Intelligence: From Secrets to Policy, 3rd edition. Washington, DC: CQ Press, 2005.
MacDonnell, Francis. Insidious Foes: the Axis Fifth Column and the American Home Front. New York: Oxford University Press, 1995.
Malia, Martin. The Soviet Tragedy: A History of Socialism in Russia, 1917-1991. NY, NY: The Free Press, 1994.
Maoz, Zeev, and Bruce Russett. “Alliances, Wealth, Contiguity and Political Stability: Is the Lack of Conflict between Democracies a Statistical Artifact?” International Interaactions 17 (1992): 245-267.
267
Maoz, Zeev, and Bruce Russett. "Normative and Structural Causes of the Democratic Peace, 1946-1986." American Political Science Review 87, no. 3 (1993): 624- 638.
Markey, Daniel. "A False Choice in Pakistan." Foreign Affairs, 2007: 85+.
Marshall, Monty G., Keith Jaggers and Ted Robert Gurr. "Polity IV Project: Political Regime Characteristics and Transitions, 1800-2010, Dataset Users' Manual." Center for Systemic Peace, 2011. http://www.systemicpeace.org/inscr/p4manualv2010.pdf (accessed June 25, 2012).
Marshall, Monty G., Ted Robert Gurr, Christian Davenport, and Keith Jaggers. "Politiy IV, 1800-1999: Comments on Munck and Verkuilen." Comparative Political Studies 35, no. 1 (February 2002): 40-45.
Martin, Lisa. "Contributions of Rational Choice: A Defense of Pluralism." International Security 23, no. 4 (1999): 74-87.
Matin, A. Michael. “The Creativity of War Planners: Armed Forces Professionals and the Pre-1914 British Invasion-Scare Genre.” ELH 60, no. 4 (2011): 801-831.
May, Earnest R. "Studying and Teaching Intelligence." Studies in Intelligence (Central Intelligence Agency) 38 (1995).
McCrummen, Stephanie. "U.S. Policy in Africa Faulted on Priorities Security Is Stressed Over Democracy." The Washington Post, February 22, 2008: A17.
McKay, Craig Graham. "British SIGINT and the Bear, 1919-1941: Some Discoveries in the GC&CS Archive." Kungl.Krigsvetenskapsakademiens handlingar och tidskrift 2 (1997): 81-96.
Mearsheimer, John J. "Back to the Future: Instability in Europe after the Cold War." International Security 15, no. 1 (1990): 5-56.
Mearsheimer, John J. "The False Promise of International Institutions." International Security 19, no. 3 (1994/1995): 5-49.
Mearsheimer, John J. The Tragedy of Great Power Politics. NY, NY: Norton & Company, 2001.
Mesquita, Bruce Bueno de, James D. Morrow, Randolph M. Siverson, and Alastair Smith. "An Institutional Explanation of the Democratic Peace." American Political Science Review 93, no. 4 (1999): 791-807.
Meyer, Josh. "Al Qaeda Grows by Adding Affiliates." Los Angeles Times, September 16, 2007: A1.
268
Miller, Gregory D. "Hypotheses on Reputation: Alliance Choices and the Shadow of the Past." Security Studies 12, no. 3 (2003): 40-78.
Mitelman, Lawrence T. "Preface to a Theory of Intelligence." CIA Historical Review Program. 1974. https://www.cia.gov/library/center-for-the-study-of- intelligence/kent-csi/docs/v18i3a03p_0001.htm (accessed March 17, 2008).
Morgan, T. Clifton, and Glenn Palmer. "To Protect and to Serve: Alliancs and Foreign Policy Portfolios." The Journal of Conflict Resolution 47, no. 2 (2003): 180-203.
Morgenthau, Hans. "Realist Theory of International Relations." In Politics Among Nations, by Hans Morgenthau, 3-15. NY, NY: Alfred A. Knopf, 1948 (1978).
Morrow, James D. “The Strategic Setting of Choices: Signaling, Commitment, and Negotiation in International Politics.” In Strategic Choice and International Relations, edited by David A. Lake and Robert Powell, 77-114. Princeton, NJ: University of Princeton Press, 1999.
Morrow, James D. "Alliances: Why Write Them Down?" Annual Review of Political Science 3, no. 1 (2000): 63-83.
Müller, Harald, and Jonas Wolff. "Dyadic Democratic Peace Strikes Back." Standing Group on International Relations. September 9, 2004. http://www.sgir.org/conference2004/papers/Mueller%20Wolff%20- %20Dyadic%20Democratic%20P eace%20Strikes%20Back.pdf. (accessed August 13, 2006).
Mulligan, Timothy P. "Spies, Ciphers and 'Zitadelle': Intelligence and the Battle of Kursk, 1943." Journal of Contemporary History 22, no. 2 (1987): 235-260.
Munck, Gerardo L., and Jay Verkuilen. "Conceptualizing and Measuring Democracy: Evaluating Alternative Indicies." Comparative Political Studies 35, no. 1 (2002): 5-34.
Munting, Roger. “Lend-Lease and the Soviet War Effort.” Journal of Contemporary History 19, no. 3 (1984): 495-510.
Munton, Don. "Our Men in Havana: Washington and Canadian Intelligence on Castro's Cuba, 1959-1963." 2002. http://circ.jmellon.com/docs/pdf/our_men_in_havana_washington_and_canadi an_intelligence_on_castros_cuba.pdf (accessed January 6, 2005).
Munton, Don. “Intelligence Cooperation Meets Internaitonal Studies Theory: Explaning Canadian Operations in Castro’s Cuba.” Intelligence and National Security 24, no. 1 (2009): 119-138.
269
National Security Archive. "Iran-Iraq War: Analysis of Possible U.S. Shift from Position of Strict Neutrality." October 7, 1983. http://www.gwu.edu/~nsarchiv/NSAEBB/NSAEBB82/iraq22.pdf (accessed September 19, 2007).
North, Douglas C., and Robert P. Thomas. The Rise of the Western World: A New Economic History. Cambridge: Cambridge University Press, 1973.
Norton-Taylor, Richard. "Right of Information Access to be Widened." The Guardian, July 16, 1993: 2.
Obama, Barack, and David Cameron. "Not just special, but an essential relationship." The Times, May 24, 2011: 17.
ODNI, Office of the Director of National Intelligence. "National Intelligence Strategy: Transformation through Integration and Innovation." October 2005. http://www.odni.gov/publications/NISOctober2005. pdf (accessed March 1, 2008).
O'Halpin, Eunan. "Small States and Big Secrets: Understanding Sigint Cooperation between Unequal Powers during the Second World War." Intelligence and National Security 17, no. 3 (2002): 1-16.
Owen, John. "When do Ideologies Produce Alliances? The Holy Roman Empire, 1517-1555." International Studies Quarterly 49 (2005): 73-99.
Payne, Stanley G. The Spanish Civil War, the Soviet Union, and Communism. New Haven, CT: Yale University Press, 2004.
Peszke, Michael Alfred. The Polish Underground Army, the Western Allies, and the Failure of Strategic Unity in World War II. Jefferson, NC: McFarland, 2005.
Pierson, Paul. Politics in Time. Princeton, NJ: Princeton University Press, 2004.
Pipes, Richard. The Russian Revolution. NY, NY: Vintage Books, 1990.
Powell, Robert. "Anarchy in International Relations Theory: The Neorealist- Neoliberal Debate." International Organization, 1994: 313-344.
Powell, Robert. "Defending against Terrorist Attacks with Limited Resources." American Political Science Review 101, no. 3 (2007): 527-541.
Przeworski, Adam. Democracy and the Market; Political and Economic Reforms in Eastern Europe and Latin America. NY, NY: Cambridge University Press, 1991.
Ramesh, Randeep, and Lee Glendinning. "Terror plot: Intelligence: Pakistan's Role in Uncovering Conspiracy." The Guardian (London), August 11, 2006: 5.
270
Ransom, Henry Howe. The Intelligence Establishment. Cambridge, MA: Cambridge University Press, 1970.
Rashid, Ahmed. "Losing the Battle for Hearts and Minds." The Daily Telegraph, September 11, 2006: 15.
Ray, James Lee. "A Lakatosian View of the Democratic Peace Research Program." Progress in International Relations. Edited by Colin Elman and Miriam Fendius Elman. Cambridge, MA: MIT Press, 2003: 205-243.
Reveron, Derek S. "Old Allies, New Friends: Intelligence-Sharing in the War on Terror." Orbis 50, no. 3 (2006): 453-468.
Reynolds, David. "Competitive Co-Operation: Anglo-American Relations in World War Two." The Historical Journal 23, no. 1 (March 1980): 233-245.
Reynolds, David. The Creation of the Anglo-American Alliance 1937-1941: A Study in Competitive Co-operation. Chapel Hill, NC: The University of North Carolina Press, 1981.
Richelson, Jeffrey T., and Desmond Ball. The Ties That Bind: Intelligence Cooperation between the UKUSA Countries, the United Kingdom, the United States of America, Canada, Australia, and New Zealand. Boston, MA: Allen & Unwin, 1985.
Richelson, Jeffrey T. "The Calculus of Intelligence Cooperation." International Journal of Intelligence and Counterintelligence 4, no. 3 (1990): 307-323.
Ripsman, Norrin M., and Jack S. Levy. "Wishful Thinking or Buying Time? The Logic of British Appeasement in the 1930s." International Security 33, no. 2 (2008): 148-181.
Risse-Kappen, Thomas. "Democratic Peace -- Warlike Democracies? A Social Constructivist Interpretation of the Liberal Argument." European Journal of International Relations 1, no. 4 (1995): 491-518.
Rohwer, Jürgen. “Review Article: The Wireless World at War, 1939-1945.” The International History Review 16, no. 3 (1994): 536-548.
Rosato, Sebastian. "The Flawed Logic of Democratic Peace Theory." American Political Science Review 97, no. 4 (2003): 585-602.
Rousseau, David L., Christopher Gelpi, Dan Reiter, and Paul Huth. "Assessing the Dyadic Nature of the Democratic Peace, 1919-88." American Political Science Review 90, no. 3 (1996): 512-533.
Rovner, Joshua, Austin Long, and Amy B. Zegart. "How Intelligent is Intelligence Reform?" International Security, 4, no. 30 (2006): 196-208.
271
Rudner, Martin. "Britain Betwixt and Between: UK SIGINT Alliance Strategy's Transatlantic and European Connections." Intelligence and National Security 19, no. 4 (2004): 571-609.
Rudner, Martin. "Canada's Communications Security Establishment from Cold War to Globalization." Intelligence and National Security 16, no. 1 (2001): 97-128.
Rummel, Rudolph J. "Libertarianism and international violence." Journal of Conflict Resolution 27, no. 1 (1983): 27-71.
Russett, Bruce. Grasping the Democratic Peace: Principles for a Post-Cold War World. Princeton, NJ: Princeton University Press, 1993.
Sagan, Scott. The Limits of Safety: Organizations, Accidents, and Nuclear Weapons. Princeton, NJ: Princeton University Press (1993).
Scheuer, Michael F. “Statement before the House Committee on Foreign Affairs, Subcommittee on International Organizations, Human Rights, and Oversight Subcommittee on Europe: Extraordinary Rendition in U.S. Counter Terrorism Policy: The Impact on Transatlantic Relations.” April 17, 2007. http://foreignaffairs.house.gov/110/sch041707.htm (accessed July 10, 2012).
Schmitter, Philippe C. "Organized Labor between the Practice of Democracy and the Prospect of Democratization of Europe." In Europe - One Labour Market?, edited by Lars Magnusson and Jan Ottosson, 179-192. Brussels: P.I.E.-Peter Lang, 2002.
Schroeder, Paul. "Historical Reality vs. Neo-Realist Theory." International Security 19, no. 1 (1994): 108-148.
Schuessler, John M. "The Deception Dividend: FDR's Undeclared War." International Security 34, no. 4 (2010): 133-165.
Schumpeter, Joseph A. "The Sociology of Imperialism." In Imperialism and Social Classes, by Joseph Schumpeter. Cleveland, IL: World Publishing Company, 1950 (1919).
Schumpeter, Joseph A. Capitalism, Socialism and Democracy. NY, NY: Harper & Row, 1976 (1942).
Scott, Len. "Sources and Methods in the Study of Intelligence: A British View." Journal of Intelligence and National Security 22, no. 2 (2007): 185-205.
Scott, Len, and Peter Jackson. "The Study of Intelligence in Theory and Practice." Journal of Intelligence and National Security 19, no. 2 (2004): 139-169.
Searle, Alaric. “Uneasy Intelligence Collaboration, Genuine Ill Will, with an Admixture of Ideology: The British Military Mission to the Soviet Union,
272
1941-45.” In Military Advising and Assistance: From Mercenaries to Privatization, 1815-2007, edited by Donald Stoker, 61-80. NY: Routledge, 2008.
Seligmann, Matthew. "Intelligence Information and the 1909 Naval Scare: The Secret Foundations of a Public Panic." War in History 17, no. 1 (2010): 37-59.
Shane, Scott. "U.S. to Declassify Secrets at Age 25." New York Times, December 21, 2006a.
Shane, Scott. "U.S. Archives Making Public Data Secret Again." The International Herald Tribune, February 22, 2006b: 2.
Shane, Scott, and Robert F. Worth. "Earlier Flight May Have Been Dry Run for Plotters." NYTimes.com. November 1, 2010. http://www.nytimes.com/2010/11/02/world/02terror.html?_r=1&pagewanted= print (accessed November 18, 2010).
Sheridan, Mary Beth, and Joby Warrick. “Joint Counterterrorism Efforts Could Face Public Opposition.” Washington Post. February 13, 2011: A1+.
Shultz, Kenneth. "Do Democratic Institutions Constrain or Inform?" International Organization 53, no. 2 (1999): 233-266.
Siegel, Eric. "I Know that You Know, and You Know that I Know: An Information Theory of the Democratic Peace." 93rd Annual Meeting of the American Political Science Association (APSA). Washington, D.C.: APSA, 1997.
Sims, Jennifer E. "Foreign Intelligence Liaison: Devils, Deals and Details." Journal of International Intelligence and Counterintelligence 19, no. 2 (2006): 195- 217.
Sims, John Cary. "The BRUSA Agreement of May 17, 1943." Cryptologia, 21, no. 1 (1997): 30-38.
Slavin, Barbara. "U.S. Support Key to Ehtiopia's Invasion; Military, Weapons Boost Tied to Ouster of Somali Interests." USA Today, January 8, 2007: 10A.
Smith, Bradley F. Sharing Secrets with Stalin: How the Allies Traded Intelligence, 1941-19456. Lawrence, KS: University of Kansas Press, 1996.
Smith, Bradley F. The Ultra-Magic Deals and the Most Secret Special Relationship, 1940-1946. Novato, CA: Presidio, 1993.
Snyder, Glenn H. "Alliance Theory: A Neo-Realist First Cut." International Affairs 44, no. 1 (1990).
273
Soybel, Phyllis L. A Necessary Relationship: the Development of Anglo-American Cooperation in Naval Intelligence. Westport, CT: Praeger, 2005.
Spence, Richard B. "Englishmen in New York: The SIS American Station, 1915-21." Intelligence and National Security 19, no. 3 (Autumn 2004): 511-537.
Spiro, David E. "The Insignificance of the Liberal Peace." International Security 19, no. 2 (1994): 50-86.
Sserwanga, Moses. "Uganda: Anti-Terrorism Legislation Abused - Not Serving Purpose for Its Being." The Monitor, April 2006, 2006.
Stafford, David, and Rhodri Jeffreys-Jones, . American-British-Canadian Intelligence Relations, 1939-2000. Portland, OR: Frank Cass, 2000.
Steele, Robert David. "Foreign Liaison and Intelligence Reform: Still in Denial." International Journal of Intelligence and Counterintelligence 20, no. 1 (2007): 167-174.
Stevenson, William. A Man Called Intrepid: The Incredible WWII Narrative of the Hero Whose Spy Network and Secret Diplomacy Changed the Course of History. NY: Ballantine, 1976.
Stewart, Richard W. American Military History: The United States Army in a Global Era, 1917-2003. Vol. II. Washington, DC, 2005.
Stinnett, Douglas M., Jaroslav Tir, Philip Schafer, Paul F. Diehl, and Charles Gochman. "The Correlates of War Project Direct Contiguity Data, Version 3." Conflict Mangagement and Peace Science 19, no. 2 (2002): 58-66.
Svendsen, Adam D.M. “Connecting Intelligence and Theory: Intelligence Liaison and International Relations.” Intelligence and National Security 24, no. 5 (2009): 700-729.
Sweeney, Kevin. "Jumping on the Bandwagon: An Interest-Based Explanation for Great Power Alliances." The Journal of Politics 66, no. 2 (2004): 428-449.
Tenet, George J. "What 'New' Role for the CIA?" The New York Times, October 27, 1998: A23.
Tenet, George J. "Written Statement for the Record of the Director of Central Intelligence Before the National Commission on Terrorist Attacks Upon the United States." Cyber Cemetary. March 24, 2004. http://govinfo.library.unt.edu/911/hearings/hearing8/tenet_statement.pdf (accessed February 15, 2008).
274
Tetlock, Philip, and Charles McGuire. "Cognitive Perspectives on Foreign Policy." In American Foreign Policy, by John G. Ikenberry, 504-516. NY, NY: Longman, 1985.
Thompson, William R. "Democracy and Peace: Putting the Cart Before the Horse?" International Organization 50, no. 1 (1996): 141-174.
Treverton, Gregory F., Seth G. Jones, Steven Boraz, and Phillip Lipscy. "Toward a Theory of Intelligence: Workshop Report." Rand Corporation. 2005. http://www.rand.org/pubs/conf_proceedings/2006/RAND_CF219.pdf (accessed August 29, 2007).
Tuzuner, Musa. The State-Level Determinants of the United States’ International Intelligence Cooperation. PhD Dissertation, Kent State University, August 2009. http://etd.ohiolink.edu/send- pdf.cgi/Tuzuner%20Musa.pdf?kent1246699828 (accessed 27 June 2012).
Ulam, Adam B. Expansion and Coexistence: Soviet Foreign Policy, 1917-1973. NY: Praeger, 1974.
U.S. News and World Report. "Best Graduate Schools: Political Science (Ranked in 2005)." U.S. News and World Report. 2005. http://gradschools.usnews.rankingandreviews.com/grad/pol/search (accessed April 1, 2008).
U.S. Senate. "The Final Report of the Select Committee To Study Governmental Operations With Respect To Intelligence Activities." The Assassination Archives and Research Center (AARC). April 26, 1976. http://www.aarclibrary.org/publib/church/reports/book1/pdf/ChurchB1_10_D omestic.pdf (accessed April 8, 2008).
Uzzi, Brian. "Embeddedness in the Making of Financial Capital: How Social Relations and Networks Benefit Firms Seeking Financing." American Sociological Review 64 (1999): 481-505.
Van Evera, Stephen. Guide to Methods for Students of Political Science. Ithaca, NY: Cornell University Press, 1997.
Von Zur Gathen, Joachim. "Zimmermann Telegram: The Original Draft." Cryptologia 31, no. 2 (2007): 2-37.
Walsh, James Igoe. "Counterinsurgency Cooperation: The Importance and Structure of Intelligence Sharing." All Academic, Inc. March 26, 2008. http://www.allacademic.com/meta/p254128_index.html (accessed April 3, 2008).
Walsh, James Igoe. "Defection and Hierarchy in International Intelligence Sharing." Journal of Public Policy 27, no. 2 (2007a): 151-181.
275
Walsh, James Igoe. "Intelligence Sharing and United States Counter-Terrorism Policy." International Security Cooperation: Intelligence Practices. Ankara, Turkey: Bilkent University, 2007b.
Walsh, James Igoe. The International Politics of Intelligence Sharing. NY, NY: Columbia University Press, 2010.
Walt, Stephen. The Origins of Alliances. Ithaca, NY: Cornell University Press, 1987.
Waltz, Kenneth. "Political Structures." In Neorealism and Its Critics, edited by Robert O. Keohane, 47-69. NY, NY: Columbia University Press, 1979 (1986).
Waltz, Kenneth. "Structural Realism After the Cold War." International Security 25, no. 1 (2000): 5-41.
Warner, Michael. "Intelligence Transformation and Intelligence Liaison." SAIS Review 24, no. 1 (2004): 77-89.
Warner, Michael. "Wanted: A Definition of Intelligence." Studies in Intelligence (Central Intelligence Agency) 46, no. 3 (2002).
Watts, Larry L. "Intelligence Reform in Europe's Emerging Democracies: Conflicting Paradigms, Dissimilar Context." Studies in Intelligence (Central Intelligence Agency) 48, no. 1 (2004).
Weber, Katja. "Hierarchy amidst Anarchy: A Transaction Costs Approach to International Security Cooperation." International Studies Quarterly 41, no. 2 (1997): 321-340.
Weber, Katja. Hierarchy Amidst Anarchy: Transaction Costs and Institutional Choice. Albany, NY: State University of New York, 2000.
Weber, Katja, and Mark Hallerberg. "Explaining variation in institutional integration in the European Union: why firms may prefer European solutions." Journal of European Public Policy 8, no. 2 (2001): 171-191.
Weber, Katja, and Michael E. Smith. "EU External Relations After Enlargement: The Complex Governance of Wider Europe." Second Pan-European Conference, Standing Group on EU Politics. Bologna, Italy: The European Consortium for Political Research, 2004.
Weber, Max. Essays in Sociology. Translated by H.H. Gerth and C. Wright Mills. Oxford: Oxford University Press, 1946.
Wendt, Alexander. "Driving with the Rearview Mirror: On the Rational Science of Institutional Design." International Organization 55, no. 4 (2001): 1019-1049.
276
West, Nigel, and Oleg Tsarev. The Crown Jewels: The British Secrets at the Heart of the KGB Archives. New Haven, CT: Yale University Press, 1999.
Wetzling, Thorsten. "The Democratic Control of Intergovernmental Intelligence Cooperation (Working Paper #165)." Geneva Centre for the Democratic Control of Armed Forces (DCAF). 2005. http://www.dcaf.ch/handbook_intelligence/Democratic_Control_Wetzling.pdf (accessed August 9, 2007).
Whitlock, Craig. "After a Decade at War with West, Al-Qaeda Still Impervious to Spies." The Washington Post, March 20, 2008: A1.
Whitlock, Craig. "Jordan's Spy Agency: Holding Cell for the CIA; Foreign Terror Suspects Tell of Torture." The Washington Post, December 1, 2007: A1.
Williamson, Oliver E. "Peak Load Pricing." American Economic Review 56 (1966): 810-827.
Williamson, Oliver E. "Transaction-Cost Economics: The Governance of Contractual Relations." Journal of Law and Economics 22, no. 2 (1979): 233-261.
Windsor, Jennifer. "Promoting Democratization Can Combat Terrorism." The Washington Quarterly 26, no. 3 (2003): 53-54.
Winterbotham, F.W. The Ultra Secret. NY, NY: Harper & Row, 1974.
Wirtz, James J. "Constraints on Intelligence Collaboration: The Domestic Dimension." Defense Analysis 8, no. 3 (1992): 247-259.
Wohlforth, William C. “The Stability of a Unipolar World.” International Security 24, no. 1 (1999): 5-41.
Wohlstetter, Roberta. Pearl Harbor: Warning and Decision. Stanford, CA: Stanford University Press, 1962.
Woodward, Bob. "CIA Aiding Iraq in Gulf War; Target Data from U.S. Satellites Supplied for Nearly 2 Years." The Washington Post, December 1986, 1986: A1+.
Young, Joseph, and Brian Urlacher. "Cantankerous Cooperation: Democracies, Authoritarian Regimes, and the Prisoner's Dilemma." International Interactions 33, no. 1 (2007): 51-73.
Young, Oran. "Regime Dynamics: the Rise and Fall of International Regimes." International Organization 36, no. 2 (1982): 277-297.
Zakaria, Fareed. "The Rise of Illiberal Democracy." Foreign Affairs 76, no. 6 (1997): 22-43.
277
Zegart, Amy B. "An Empirical Analysis of Failed Intelligence Reform Before September 11." Political Science Quarterly 121, no. 1 (2006): 33-60.
Zegart, Amy B. "September 11 and the Adaptation Failure of U.S. Intelligence Agencies." International Security 29, no. 4 (2005): 78-111.
Zegart, Amy B. Spying Blind: The CIA, the FBI, and the Origins of 9/11. Princeton, NJ: Princeton University Press, 2007.
Zegart, Amy B. Eyes on Spies: Congress and the United States Intelligence Community. Stanford, CA: Hoover Institution Press, 2011.
Zürn, Michael. “The Rise of International Environmental Politics: A Review of Current Research.” World Politics 50, no. 4 (1998): 617-649.
278